Query filtering in the ingester and storage

[MasslessParticle]

This PR introduces the concept of a FilteringPipeline/FilteringSampleExtractor that take 0 or more PipelineFilters. PipelineFilters are just log pipelines that run before the Pipeline/Extractor created by the query. Any PipelineFilter that returns true is omitted. The FilteringPipeline is then passed to the underlying iterators like any other pipeline.

Another notable change is that this PR changed the Pipeline interface to also accept timeranges. This is to accommodate the deletes that only partially affect the queried timerange.

The original POC for this work had ChunkFilters but it turns out they're not needed because a delete filter needs to match on labels anyway.

There's a potential future improvement where we use a chunk filter when chunks match a delete's labels and are totally within its timerange.

This PR will change Querying behavior in installations with current delete requests so we should wait until
#5481 is merged because it contains a feature flag for delete behavior

Checklist

• Documentation added
• Tests updated
• Add an entry in the CHANGELOG.md about the changes.

[MichelHollands]

LGTM. A few nit comments were added.

[sandeepsukhani]

I found some minor issues but overall the code looks great to me.
Please let me know if you need any more inputs on any of my comments.

[MasslessParticle]

I've addressed nearly all the feedback. I still need some help from @sandeepsukhani to understand the filtering. I've added some tests cases to the filtering and everything behaves as I expect so I'm not sure what I'm missing.

The tests are as follows:

2 filters:

• filter 1: {foo="bar", bar="baz"} | = e
• filter 2: {baz="foo"} |= e

Cases (not including timerange stuff):

• Stream is {foo="bar"}: ok = true because no filter matches this stream
• Stream is {beep="boop"}: ok = true because no filter matches this stream
• Stream is {foo="bar", bar="baz"}: ok = false because filter 1 matches
• Stream is {baz="foo"}: ok = true because filter 2 matches

[cyriltovena]

Looking good let's make sure the allMatch is good.

[MasslessParticle]

All match is good and I've incorporated @sandeepsukhani 's feedback

[sandeepsukhani]

Suggested some very minor nits but other than that it LGTM! Nice work!