Add redis and inmemory cache backend for blocks storage

[kamijin-fanta]

What this PR does:

• add Redis cache backend
• add configure options
  • -blocks-storage.bucket-store.index-cache.backend=redis
  • -blocks-storage.bucket-store.index-cache.redis.*
  • -blocks-storage.bucket-store.metadata-cache.backend=redis
  • -blocks-storage.bucket-store.metadata-cache.redis.*
  • -blocks-storage.bucket-store.chunks-cache.backend.backend=redis
  • -blocks-storage.bucket-store.chunks-cache.backend.redis.*
  • -blocks-storage.bucket-store.metadata-cache.inmemory.*
  • -blocks-storage.bucket-store.chunks-cache.inmemory.*

Which issue(s) this PR fixes:
fix #3568

Checklist

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

[kamijin-fanta]

I have some questions.

• There are several variations of RedisConfig. Please let me know if there is a reference implementation of RegisterFlagsWithPrefix.
  • The TTL is passed from thanos to the metadata and chunks cache. -blocks-storage.bucket-store.[metadata-cache|chunks-cache].redis.expiration is no longer respected. I want to remove this.
  • The index cache wants to set the default expiration to 24h. However, it is currently set to 0.
• How should I set the Prometheus metric name of RedisIndexCache?
  • The current code uses the same thanos_store_index_cache_requests_total as Thanos. I think there is a possibility of using the cortex namespace.
• Are there any tests I should add?

[pstibrany]

I believe there is an ongoing work in Thanos to move IndexCache to use cache.Cache interface. I would wait until it lands, to avoid copying Thanos' MemcachedIndexCache here.

(I cannot find an issue for it though. I think it is being done by @Sudhar287 as part of thanos-io/thanos#3408, but I may be confusing the two, and it may in fact be unrelated.)

[kamijin-fanta]

Since the related PR is closed, work will resume. I also added an InMemory implementation to the cache.

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 60 days. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[alanprot]

I think this is an interesting option. Are we going to go forward with this?

[phin1x]

@kamijin-fanta @pstibrany @pracucci can we re-open this pr? redis is also very popular and should be an alternative besides memcached.