MDEV-20228 `mysql_upgrade` fails on every version upgrade: "ERROR 1267 (HY000) at line 7: Illegal mix of collations (utf8mb4_unicode_ci,COERCIBLE) and (utf8mb4_general_ci,COERCIBLE) for operation 'like'"

MDEV-18153 Assertion `0' or Assertion `btr_validate_index(index, 0)' failed in row_upd_sec_index_entry or error code 126: Index is corrupted upon UPDATE with TIME_ROUND_FRACTIONAL

Conversion to a temporal data type resulting into a lower precision
depends on TIME_ROUND_FRACTIONAL. Taking into account this dependency in:
- indexed generated virtual column expressions
- persistent virtual column expressions

A warning is now issued if conversion from the generation expression
to the column data type depends on TIME_ROUND_FRACTIONAL.

The warning will be changed to error in 10.5

Followup patch: fix typos

innobase_rename_or_enlarge_column_try(): Relax a debug assertion.
The DATA_VERSIONED bits in dict_col_t::prtype do not affect the
storage format.

Fixed 4-byte length characters handled incorrectly.

Problem was that if user used bad wsrep configuration
we call unireg_abort and in wsrep case there is check
to wsrep_server_state that might not yet be initialized.
Fixed by checking that wsrep_server_state is initialized
before checking it's state.

Changes to be committed:
	modified:   sql/mysqld.cc
	modified:   sql/wsrep_server_state.h

In mysql-server/commit@f46329044f8618212923bdf52e15d5b464201edc
the InnoDB function btr_cur_open_at_rnd_pos() was corrected so that
it would return a status that indicates whether the cursor was
successfully positioned. But this change was not correctly merged to
MariaDB in 2e814d47.

btr_cur_open_at_rnd_pos(): In the code path that was introduced in
MDEV-8588, properly return failure status.

No deterministic test case was found for this failure.
It was caught after removing the function
page_copy_rec_list_end_to_created_page() in a development branch.
As a result, the fill factor of index trees would improve, and
supposedly, so would the probability of btr_cur_open_at_rnd_pos()
reaching the intentionally corrupted page in the test
innodb.leaf_page_corrupted_during_recovery.
The wrong return value would cause
btr_estimate_number_of_different_key_vals() to wrongly invoke
btr_rec_get_externally_stored_len() on a non-leaf page and
trigger an assertion failure at the start of that function.

- During trx_undo_report_rename(), InnoDB can fail to write undo log
for it if undo log doesn't fit in the undo page. In that case, InnoDB
adds one more undo log page and retry to write the rename undo log.
But the assert is wrong and it doesn't allow to fail even for one time.

Null merge the MDEV-14383 fix; it affects 5.5 only.

Also added comments for trace_plan_prefix and the class Json_writer_temp_disable

Added:
      1) estimated_join_cardinality
      2) best_chosen_access_method for a table
      3) best_join_order

there was a SELECT/DELETE race

Files for PAGE_COMPRESSED tables that were created with
innodb_checksum_algorithm=full_crc32 store the value of
innodb_compression_algorithm at the time of the file creation.

The server-wide setting of innodb_compression_algorithm
may be changed after file creation. We must ignore any mismatch
when opening a data file, and for writes, we must use the
choice of algorithm that is stored in the file.

fil_space_t::is_flags_full_crc32_equal(): Ignore the
innodb_compression_algorithm but do compare innodb_page_size.

fil_space_t::is_flags_non_full_crc32_equal(): Ignore the
innodb_compression_algorithm.

The test mariabackup.encrypted_page_corruption was hard-wired
for innodb_checksum_algorithm=crc32. Enable a combination for
full_crc32 for the sake of completeness.

Re-enable some Galera tests that should have been enabled.

Add client_ed25519.so to debian/libmariadb3.install;
merge e47a143f correctly.

Remove a duplicated #include from wsrep_mysqld.cc.

* do not allow versioned table to be without versioned (non-system) fields
* prohibit changing field versioning, when removing table versioning
* handle CREATE...SELECT as well

The setting innodb_change_buffering_debug=2 was supposed to inject
a crash during change buffer merge. There is no public test for
that functionality, and even if there were, it would be better
to use DEBUG_SYNC to halt the thread that does change buffer merge,
force a redo log flush from another thread, and finally kill the
server externally.

The most likely cause of the crash is that a timer fired, after it was closed.

MSDN documents such a possibility, in the documentation for
CloseThreadpoolTimer() function, and recommends disabling the timer before
calling WaitForThreadpoolTimerCallbacks()/CloseThreadpoolTimer().

The fix follows this recommendation.

Note, that 5.5-10.1 disabled the timer before close, but this code
was lost in threadpool refactoring in 10.2

MYSQL_MAINTAINER_MODE is ERR