Implement grpc.Compressor.DecompressedSize for snappy to optimize memory allocations

[damnever]

CPU usage can be reduced by up to 25% with this optimization.

Checklist

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

[alanprot]

This is nice!!! I wanted to do this before as well!! :D

Did u see any improvements on ingesters CPU?

[damnever]

Did u see any improvements on ingesters CPU?

@alanprot In our test environment, without heavy queries, the CPU usage can be reduced by up to 25%, roughly 10% on average.

[alanprot]

LGTM. Thanks

[alanprot]

Ok..

Seems that something weird here: DecodedLen does not return the size of the uncompressed data but rather the size of the decoded block.

I was writing unit tests for this and i noticed that:

ex:

				buf = bytes.NewBuffer(compressedBytes)
				r, err := c.Decompress(buf)
				require.NoError(t, err)
				out, err := io.ReadAll(r)
				assert.Equal(t, len(out), sizer.DecompressedSize(compressedBytes))

Expected :18216
Actual   :895

I think we cannot use decodeLen when using stream. From https://pkg.go.dev/github.com/klauspost/compress/snappy#section-readme

There are actually two Snappy formats: block and stream. They are related, but different: trying to decompress block-compressed data as a Snappy stream will fail, and vice versa. The block format is the Decode and Encode functions and the stream format is the Reader and Writer types.