monitoring: apm-server not shiping tbs monitoring metrics

[endorama]

Follow up on #13475

Not all metrics have been included in the previous fixes. One of which is beat.stats.apm_server.sampling.tail.*.

### Tasks
- [ ] https://github.com/elastic/elasticsearch/pull/121543
- [ ] https://github.com/elastic/beats/pull/42569
- [ ] https://github.com/elastic/integrations/pull/12568

How to test these PRs

See #13475

Enable TBS, see that metrics are indexed.

[lucabelluccini]

If possible, let's also publish:

• the value of sampling.tail.storage_limit as a number so we can easily plot it
• a boolean metric to tell if TBS is "bypassed" because of the sampling.tail.storage_limit is exceeded (instead of relying on logs)

[carsonip]

the value of sampling.tail.storage_limit as a number so we can easily plot it

Created #15533 to track this

a boolean metric to tell if TBS is "bypassed" because of the sampling.tail.storage_limit is exceeded (instead of relying on logs)

It doesn't have to be a boolean metric. I believe the existing sampling.tail.events.failed_writes should be good enough to determine this. It should be available once this issue is resolved.