Fix reuse slice in distributor.Push()

[pracucci]

What this PR does:
While investigating the issue #1895, I've realized the request slice is not reused in case the request gets skipped because of HA dedup or rate limiting. In case all incoming requests have the HA enabled, it's 50% of requests. This PR fixes it.

Which issue(s) this PR fixes:
No issue

Checklist

• Tests updated
• Documentation added
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]

[codesome]

I am wondering if we should do it in a defer statement if there was an error?

[pracucci]

I am wondering if we should do it in a defer statement if there was an error?

@codesome In case of error it's OK doing it before return, so I don't think there's any real value paying the cost of a defer. If you mean adding a defer at the beginning of the function, to cover all cases at once, unfortunately we can't do it because of the async nature of ring.DoBatch(): when the execution reaches that point, the req needs to be alive even after the distributor.Push() completes.