Make search case sensitive

[mdisibio]

What this PR does:
This PR makes all search-related activity (including tag lookups) case-sensitive. An original intent of this was to reduce cardinality and resource usage of ingester-search, which is extra data stored next to the actual block. Therefore this change will increase usage some, but the amount depends on the traffic. Parquet backend search will be more efficient due to the removal of ToLower. v2 is unaffected (was already case-sensitive).

An increase in resources at the ingester is likely to cause more occurrence of #1258 so there is an attempt to fix it. We weren't able to conclusively determine the cause of such enormous buffers, but it seems related to tags that are very large. Changes:

1. Only store the first 1024 bytes of tags recorded in flatbuffers. This loses some search-ability, but seems like a worthwhile trade off to keep data sizes down.
2. There is a soft cap of ~900MB on the size of buffers generated in tempofb. After hitting that limit it will stop writing further tags, and any remaining (ignored) data won't be searchable. The rationale for this number is explained in the comments, but I want to point out that this is still an enormous buffer, and unlikely to work well, but at least it won't panic, and will make further investigation easier.

Which issue(s) this PR fixes:
Fixes #1547 #1258

Checklist

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