[Bugfix] Deadlock on invalid query to api/v2/search/tags

[ruslan-mikhailov]

What this PR does

It fixes deadlock in ingesters.

1. panic in ExecuteTagNames on nil pointer dereference, when rootExpr is nil, which can happen when query is invalid
2. after recovering from panic, mutex unlock is not defered, which results in deadlock.

Impact: deadlock for read and write until ingester's restart for tenant who triggered panicked request to api/v2/search/tags

How to reproduce

1. send request to api/v2/search/tags with invalid query. Example:

--> GET http://localhost:3100/api/v2/search/tags
--> query params: {'q': '{ a="test" }'}
<-- status: 500
<-- body: error querying ingesters in Querier.SearchTags: failed to execute f() for x.x.x.x:x: rpc error: code = Unknown desc = recovered in SearchTagsV2

4. Try to search by trace id, it will always return 500. Example:

--> GET http://localhost:3100/api/traces/19a8654f9aaa456b7e45d19ea770ac03
<-- status: 500
<-- body: error querying ingesters in Querier.FindTraceByID: context deadline exceeded

Which issue(s) this PR fixes

Fixes #

Checklist

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