* Add server-side session watchdog

* Add test

* Remove whitespace

* Move watchdog to uaprocessor

* Restore import order

* Resolve timezone deprecation warnings

* Update uaprotocol_auto

* Fix epoch to UTC

* More UTC fixes

* Require newer cryptography

Since not_valid_before_utc and not_valid_after_utc were added

* Fix import

* Fix tz in test_datetime

* Fix ruff check

* One more import

* Include timezone in env

* Fix typo

Bumps [pypa/gh-action-pypi-publish](https://github.com/pypa/gh-action-pypi-publish) from 1.8.14 to 1.9.0.
- [Release notes](https://github.com/pypa/gh-action-pypi-publish/releases)
- [Commits](https://github.com/pypa/gh-action-pypi-publish/compare/81e9d935c883d0b210363ab89cf05f3894778450...ec4db0b4ddc65acdf4bff5fa45ac92d78b56bdf0)

---
updated-dependencies:
- dependency-name: pypa/gh-action-pypi-publish
  dependency-type: direct:production
  update-type: version-update:semver-minor
...
Signed-off-by: dependabot[bot] <support@github.com>

* Add callback to delete subscriptions

* Improve docstring

* Don't shadow CancelledError

* Simplify subscription storage

* Make it safe to iterate over keys

* Restore accidentally deleted line

* Remove task context

* Remove list

* Fix subscriptions for sessions

Fix SecurityPolicyType docstring: e.g. "None" -> "NoSecurity"
Reorganize imports and better align code to pep standards (ruff)

Bumps [actions/checkout](https://github.com/actions/checkout) from 2 to 4.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/v2...v4)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-major
...
Signed-off-by: dependabot[bot] <support@github.com>

Bumps [pypa/gh-action-pypi-publish](https://github.com/pypa/gh-action-pypi-publish) from 1.4.2 to 1.8.14.
- [Release notes](https://github.com/pypa/gh-action-pypi-publish/releases)
- [Commits](https://github.com/pypa/gh-action-pypi-publish/compare/27b31702a0e7fc50959f5ad993c78deac1bdfc29...81e9d935c883d0b210363ab89cf05f3894778450)

---
updated-dependencies:
- dependency-name: pypa/gh-action-pypi-publish
  dependency-type: direct:production
  update-type: version-update:semver-minor
...
Signed-off-by: dependabot[bot] <support@github.com>

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 2 to 3.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/v2...v3)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-major
...
Signed-off-by: dependabot[bot] <support@github.com>

Bumps [actions/setup-python](https://github.com/actions/setup-python) from 3 to 5.
- [Release notes](https://github.com/actions/setup-python/releases)
- [Commits](https://github.com/actions/setup-python/compare/v3...v5)

---
updated-dependencies:
- dependency-name: actions/setup-python
  dependency-type: direct:production
  update-type: version-update:semver-major
...
Signed-off-by: dependabot[bot] <support@github.com>

To the `import_xml` an optional argument `auto_load_definitions` is added. Default is True.

Case:
With commit de8269e4 `import_xml` automatic load(on the fly code generation) the type defintions.

When using your own extension object and enum implementation (with `ua.register_extension_object`  `ua.register_enum`) this is unwanted behaviour.

Previous work flow was:
* import xml
* register own extension object and enum implementations
* for the remain stuff call `load_data_type_definitions()`

To make this possible again an option is required to prevent the auto load definitions when calling the `import_xml` method.

in favor of their new pendants in the lint-section.

This gets rid of warnings in ruffs execution.

in logging statements (treewide)

This completes @schroeders- effort in
f48b4f51: 'remove f-string in logging '

Generally this is more consistent, more efficient, and less error-prone than formatting the string directly.

These tests should have asserted that the semicolon was actually part of the resulting NodeId as per OPC-UA spec, because string identifiers may contain arbitrary characters except Unicode control characters (e.g. semicolons).

Example:
The node `ns=9;s=Line1.nsuri=MACHINE.NS;s=MACHINE.NS.State.Running` is regularly deconstructed as NamespaceId `9`, Identifier Type `String`, Identifier `Line1.nsuri=MACHINE.NS;s=MACHINE.NS.State.Running`. Previously this was parsed to have the Identifier `MACHINE.NS.State.Running`

Mark it as expexted to fail similar to the early return contained before.

in internal_server and address_space.
Path.is_file does not consume strings,
as shelf.open does only consume string and not pathlib.Path.

to clarify its intended usage.

This confirms @harriv's finding in #1313.

I noticed that raising a `BadOutOfRangeError` in a server callback set
through `Server.subscribe_server_callback` resulted in a
*BadInternalError* on the client side (tested with UaExpert) rather than
the expected *BadOutOfRange*. But raising a
`ServiceError(ua.StatusCodes.BadOutOfRange)` instead would show the the
expected *BadOutOfRange* error in the client.
The reason for this idiosyncrasy was that `UaStatusCodeError`s were not
explicitly caught. Therefore the "default" `except Exception` handler
would be used which resulted in a
`ua.StatusCode(ua.StatusCodes.BadInternalError)` to be sent to the
client instead of the actual status code of the error that occurred.

This commit changes the behavior such that the status code of
`UaStatusCodeError`s is respected the same way as it's already done for
`ServiceError`s so that the client correctly receives the status code of
the `raise`d error.

See also https://github.com/FreeOpcUa/opcua-asyncio/discussions/1563#discussioncomment-8382555

This method allows to set a custom setter function on a per-attribute
basis. This setter function will then be called by
`write_attribute_value()` instead of writing the attribute value
directly. This allows for custom logic to be implemented by the user
(such as checking if the value to be written is within a certain range)
and if not, an appropriate error can be raised from that setter,
preventing the write and notifying the client about the occurred error.