[REF] Refactor ACL Contact Cache generation to be more efficient

[seamuslee001]

Overview

This picks up the ACL contact Cache changes from Matt's patches here https://github.com/mattwire/civicrm-groupaclcache/blob/master/patches/master...mattwire:refactor_groupaclcache.diff#L4 but uses the standardised temp table class to generate the temporary table. AUG have been running with this for the last couple of days and haven't seen any performance degradation or any issues around access control so far.

Before

Cache building is less efficient.

After

Cache building is more efficient

Technical Details

A couple of main differences here is

1. we add in a locking to ensure multiple processes don't try and fill the cache for the specific user at the same time
2. We use a temporary table but don't do a left JOIN onto civicrm_acl_contact_cache as that causes some performance issues and also is mostly notionally redundant by virtue of the check in L225 -> L237 which essentially checks to see if there are any rows in the acl contact cache for this user for that particular operation (view or edit) and aborts if there are

ping @eileenmcnaughton @mattwire @monishdeb @pfigel @aydun

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[mattwire]

@seamuslee001 So do you think the changes in this file can be merged separate to the other changes in #15588?

[eileenmcnaughton]

I would endorse merging this if others have tested / are OK with it. I think it makes the mostt important change from @mattwire's work - ie separating temp table creation from insert into possibly congested table

[mattwire]

Merging per comments in mattermost + long-term use some of my client sites