Update SVG sanitizer to conform to SVG 1.1

Original SVG sanitizer would strip out necessary elements and attributes.

Use a custom Loofah scrubber since sanitize 2.x transformers are inadequate to handle case-sensitive SVG attributes since they parse documents as HTML instead of XML, which causes all SVG attribute names (e.g. `viewBox`) to be downcased.

* SVG element list: https://www.w3.org/TR/SVG/eltindex.html
* SVG attribute list: https://www.w3.org/TR/SVG/attindex.html

Closes #14555

See merge request !3401

Reduce delay in destroying a project from 1-minute to immediately

This 1-minute delay was causing confusion among customers. Was there a reason this needed to be 1 minute?


See merge request !3997

Run ProjectDestroyWorker after pending_delete attribute has been committed to DB

Use a custom Loofah scrubber since sanitize 2.x transformers are inadequate
to handle case-sensitive SVG attributes. sanitize parses documents as HTML
instead of XML, which causes all SVG attribute names (e.g. viewBox) to be downcased.

* SVG element list: https://www.w3.org/TR/SVG/eltindex.html
* SVG attribute list: https://www.w3.org/TR/SVG/attindex.html

Closes #14555

Make group settings button white instead of gray for better visibility
Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

cc @annabeldunstone @skyruler 

See merge request !4078

Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

Add nice new Todo bell

![H5bp___Html5_Boilerplate___GitLab](/uploads/e8725396a5d77a8db09742f977595605/H5bp___Html5_Boilerplate___GitLab.png)

![H5bp___Html5_Boilerplate___GitLab](/uploads/011a21ee1653502c4f78543e28d97b67/H5bp___Html5_Boilerplate___GitLab.png)

![H5bp___Html5_Boilerplate___GitLab](/uploads/0858cb9621463ef6a32237fcb0b4f63d/H5bp___Html5_Boilerplate___GitLab.png)

Fixes: #15671 

See merge request !4077

Move group settings navigation to own partial

A bit of refactoring for !3980
Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

See merge request !4076

Align code within links in comments

Closes #17343

<img src="/uploads/78e03c1dc0e99c7bdb686d9d120a4525/Screen_Shot_2016-05-06_at_9.28.26_AM.png" width="800px">

See merge request !4073

Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

Annotate models

See merge request !4071

Update db/schema.rb after running db:migrate:reset

See merge request !4074

Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

Signed-off-by: Dmitriy Zaporozhets <dmitriy.zaporozhets@gmail.com>

Redesign navigation for group pages

Part of #14838. This MR targets on desktop version. Mobile version improvements will be in separate merge request 

See merge request !3980

Auto-set title for branches created from issues

This sets the title for a new MR to 'Resolves "$issue-title"' when:
- The source branch for the MR begins with a value iid.
- The MR has more than one commit in its diff (if there's one commit, keep using the commit's first line).
- The iid does not point to a confidential issue.

Single commit:

![A single commit uses the commit title](/uploads/cd34f59cd67f095c3034fae07950f8b5/image.png)

Multiple commits:

![Multiple commits use the issue title](/uploads/a322c406ddd56913c5aebd88d16e5a5e/image.png)

Confidential issue:

![A confidential issue uses the branch name](/uploads/7ae9b79de5f6101ced46802f3c3a6e71/image.png)

cc @DouweM @zj 

Closes #14564

See merge request !3966

Use outer join for issues ordering by milestones due.

This MR contains an implementation for Solution 1 suggested in #14183 

Ordering by Milestone due sooner/later ("milestone sort") filters issues without a milestone, which can let users feel like disappearing issues without a milestone.

See merge request !3872

If a branch starts with an issue's IID, followed by a hyphen, the
description will be updated to say that is closes the issue. This also
updates the title of the merge request to 'Resolves "$issue-title"', as
long as:
- There is more than one commit in the merge request (if there is only
  one commit, the commit's title will be used as before)
- The issue's IID is valid for the project

Update annotations on models



See merge request !3962

Assign labels milestone when moving issue

Closes #14532.

See merge request !3934

Add note to requirements doc on swap recommendation, and ensuring enough available memory.

Called out as a result of 2GB RAM DO droplets not being enough to run backups on without adding a swap file. See: https://gitlab.com/gitlab-org/omnibus-gitlab/issues/1267

cc\ @stanhu @axil 

See merge request !4060

clarify changelog entry on GitHub Enterprise import support

/cc @stanhu 

See merge request !4066

[ci skip]

Rake drop tables with cascade

See merge request !4020

Fix importer bug when throwing exceptions

Fixes https://gitlab.com/gitlab-org/gitlab-ce/issues/15681

See merge request !3941

Add a note to testing docs about the `:each` argument to RSpec hooks

[ci skip]

See merge request !4061

[ci skip]

Use tag_exists? in GitAccess#protected_tag?

This will fail until https://gitlab.com/gitlab-org/gitlab_git/merge_requests/81/ is merged. Adding this should speed up checking for tags in `Gitlab::GitAccess` as it no longer needs to retrieve the entire list of tags.

See https://gitlab.com/gitlab-org/gitlab-ce/issues/14280 for more information.

See merge request !3983

Previously this method would directly receive the output of tag_name().
This method could either return a String or nil. In the previous setup
this would somehow magically work but because Rugged::TagCollection#[]
doesn't accept nil values it started to fail.

To work around this the elsif in change_access_check() assigns the
result of tag_name() to a local and then _only_ calls protected_tag?()
if the tag name is not nil. The extra parenthesis are put in place to
ensure that things are parsed correctly, without these the code would be
parsed as follows:

    elsif tag_ref = (tag_name(ref) && protected_tag(tag_ref))

During runtime this would basically resolve to:

    elsif tag_ref = (tag_name(ref) && protected_tag(nil))

This is because when you refer to the variable you're assigning _in_ the
assignment Ruby returns nil instead of raising an error.

This removes the need for retrieving the entire list of tags just to
check if a specific one exists.