If a site specifies a relative URL root, emoji files would omit the path from
the URL, leading to lots of 404s.

Closes #15642

This stands as an alternative to using OAuth to access a user's Github
repositories.  This is setup in such a way that it can be used without OAuth
configuration.

From a UI perspective, the how to import modal has been replaced by a full
page, which includes a form for posting a personal access token back to the
Import::GithubController.

If the user has logged in via GitHub, skip the Personal Access Token and go
directly to Github for an access token via OAuth.

When pushing commits to existing branches we don’t
need to flush branch git data (branch names / counts)

When flushes the cache when pushing commits skip to
flush branch and tag git data (names / counts) because
those operations are managed explicitly in each case

Repopulated expired cache as soon as possible

IssueReferenceFilter will end up processing internal issue references
when a project uses an external issues tracker while still using
internal issue references (in the form of `#\d+`). This commit ensures
that these links are rendered as external issue links, regardless of
whether the project one currently views uses an internal or external
issues tracker.

Fixes gitlab-org/gitlab-ce#19036, gitlab-com/performance#16

This method now should return false if build is not completed. If build
is running then it is not retryable, therefore `Ci::Build#retryable?`
returned wrong value. This commit changes this behavior

This allows us to control the trusted proxies while deployed in a private network. Normally Rack::Request will trust all private IPs as trusted proxies, which can caue problems if your users are connection on you network via private IP ranges.

Normally in a rails app this is handled by action_dispatch request, but rack_attack is specifically using the Rack::Request object instead.

If Redis were not running or USE_DB were set to false, the
application settings retrieval would fail completely. This
change only attempts to use the cache if the system actually
wants to connect to the DB and rescues any failures in talking to
Redis.

Closes #17557

In #19273, we saw that retrieving ProjectTeam#human_max_access for each
note takes the bulk of the time when rendering certain issues or merge requests.
We observe that most of the comments in an issue are typically done by the
same users. This MR memoizes the max access level by user ID.

Requested in #14669

Added note to changelog

Process.clock_gettime allows getting the real time in nanoseconds as
well as allowing one to get a monotonic timestamp. This offers greater
accuracy without the overhead of having to allocate a Time instance. In
general using Time.now/Time.new is about 2x slower than using
Process.clock_gettime(). For example:

    require 'benchmark/ips'

    Benchmark.ips do |bench|
      bench.report 'Time.now' do
        Time.now.to_f
      end

      bench.report 'clock_gettime' do
        Process.clock_gettime(Process::CLOCK_MONOTONIC, :millisecond)
      end

      bench.compare!
    end

Running this benchmark gives:

    Calculating -------------------------------------
                Time.now   108.052k i/100ms
           clock_gettime   125.984k i/100ms
    -------------------------------------------------
                Time.now      2.343M (± 7.1%) i/s -     11.670M
           clock_gettime      4.979M (± 0.8%) i/s -     24.945M

    Comparison:
           clock_gettime:  4979393.8 i/s
                Time.now:  2342986.8 i/s - 2.13x slower

Another benefit of using Process.clock_gettime() is that we can simplify
the code a bit since it can give timestamps in nanoseconds out of the
box.

Depends on the changes in Workhorse (gitlab-org/gitlab-workhorse!48).

Feedback:
https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/4961#note_12738607