Re-organize schema. Drop protected from runner. Add access_level to runner. Drop protected from pipeline. Add protected to ci_bilds.

Solution 1. Persists protected(ref) flag on ci_pipelines table. builds_for_shared_runner and builds_for_specific_runner read the flag instead of executing protected_for? one by one.

`allowed_key_types` is removed and the `minimum_<type>_bits` fields are
renamed to `<tech>_key_restriction`. A special sentinel value (`-1`) signifies
that the key type is disabled.

This also feeds through to the UI - checkboxes per key type are out, inline
selection of "forbidden" and "allowed" (i.e., no restrictions) are in.

As with the previous model, unknown key types are disallowed, even if the
underlying ssh daemon happens to support them. The defaults have also been
changed from the lowest known bit size to "no restriction". So if someone
does happen to have a 768-bit RSA key, it will continue to work on upgrade, at
least until the administrator restricts them.

This is an amalgamation of:

* Cory Hinshaw: Initial implementation !5552
* Rémy Coutable: Updates !9350
* Nick Thomas: Resolve conflicts and add ED25519 support !13712

This column isn't always set (e.g. when upgrading from older instances)
and technically it could be NULL (e.g. when flushing the cache).

Fixes https://gitlab.com/gitlab-org/gitlab-ce/issues/36919

Due to a missing `on_delete: :cascade`, users would hit the error that
looked like:

```
PG::ForeignKeyViolation: ERROR: update or delete on table "protected_tags"
violates foreign key constraint "fk_rails_f7dfda8c51" on table
"protected_tag_create_access_levels" DETAIL: Key (id)=(1385) is still
referenced from table "protected_tag_create_access_levels". : DELETE FROM
"protected_tags" WHERE "protected_tags"."id" = 1385
```

Closes #36013

Caching of BroadcastMessage instances has been changed so a cache stays
valid as long as the default cache expiration time permits, instead of
the cache being expired after 1 minute. When modifying broadcast
messages the cache is flushed automatically.

To remove the need for performing sequence scans on the
"broadcast_messages" table we also add an index on (starts_at, ends_at,
id), permitting PostgreSQL to use an index scan to get all necessary
data.

Finally this commit adds a few NOT NULL constraints to the table to
match the Rails validations.

Fixes gitlab-org/gitlab-ce#31706

This changes various controllers to use the new EventCollection class
for retrieving events. This class uses a JOIN LATERAL query on
PostgreSQL to retrieve queries in a more efficient way, while falling
back to a simpler / less efficient query for MySQL.

The EventCollection class also includes a limit on the number of events
to display to prevent malicious users from cycling through all events,
as doing so could put a lot of pressure on the database.

JOIN LATERAL is only supported on PostgreSQL starting with version 9.3.0
and as such this optimisation is only used when using PostgreSQL 9.3 or
newer.

This commit migrates events data in such a way that push events are
stored much more efficiently. This is done by creating a shadow table
called "events_for_migration", and a table called "push_event_payloads"
which is used for storing push data of push events. The background
migration in this commit will copy events from the "events" table into
the "events_for_migration" table, push events in will also have a row
created in "push_event_payloads".

This approach allows us to reclaim space in the next release by simply
swapping the "events" and "events_for_migration" tables, then dropping
the old events (now "events_for_migration") table.

The new table structure is also optimised for storage space, and does
not include the unused "title" column nor the "data" column (since this
data is moved to "push_event_payloads").

== Newly Created Events

Newly created events are inserted into both "events" and
"events_for_migration", both using the exact same primary key value. The
table "push_event_payloads" in turn has a foreign key to the _shadow_
table. This removes the need for recreating and validating the foreign
key after swapping the tables. Since the shadow table also has a foreign
key to "projects.id" we also don't have to worry about orphaned rows.

This approach however does require some additional storage as we're
duplicating a portion of the events data for at least 1 release. The
exact amount is hard to estimate, but for GitLab.com this is expected to
be between 10 and 20 GB at most. The background migration in this commit
deliberately does _not_ update the "events" table as doing so would put
a lot of pressure on PostgreSQL's auto vacuuming system.

== Supporting Both Old And New Events

Application code has also been adjusted to support push events using
both the old and new data formats. This is done by creating a PushEvent
class which extends the regular Event class. Using Rails' Single Table
Inheritance system we can ensure the right class is used for the right
data, which in this case is based on the value of `events.action`. To
support displaying old and new data at the same time the PushEvent class
re-defines a few methods of the Event class, falling back to their
original implementations for push events in the old format.

Once all existing events have been migrated the various push event
related methods can be removed from the Event model, and the calls to
`super` can be removed from the methods in the PushEvent model.

The UI and event atom feed have also been slightly changed to better
handle this new setup, fortunately only a few changes were necessary to
make this work.

== API Changes

The API only displays push data of events in the new format. Supporting
both formats in the API is a bit more difficult compared to the UI.
Since the old push data was not really well documented (apart from one
example that used an incorrect "action" nmae) I decided that supporting
both was not worth the effort, especially since events will be migrated
in a few days _and_ new events are created in the correct format.

This caches the result of Appearance.first in a similar fashion to how
ApplicationSetting instances are cached. We also add some NOT NULL
constraints to the table and correct the timestamp types.

Fixes gitlab-org/gitlab-ce#36066, fixes gitlab-org/gitlab-ce#31698

By adding various composite indexes we can reduce the time spent
retrieving issue lists. Because of the way these indexes are built
column wise we can also remove some standalone indexes, keeping the
total number of indexes in check.

This was migrated to separate columns in 9.4, and now just needs to be removed
for real.

also:

* reorder table columns
* no need for `add_concurrent_index`
* no need for explicit index removal on `#down`

Previously, we used Psych, which would:

1. Check if a string was encoded as binary, and not ASCII-compatible.
2. Add the !binary tag in that case.
3. Convert to base64.

We need to do the same thing, using a new column in place of the tag.