feat(server): Add support for PFADD and PFCOUNT

[chakaz]

This version does not create sparse-encoded HLLs, however it is fully compatible with such ones created by Redis as it converts them to the dense encoding.

#925

[romange]

Wow, super cool!

[chakaz]

It should be enough to check just after inserts, as it doesn't mutate between reads

It's true that all HLLs created by Dragonfly should always be in the dense format, but in order to be compatible with Redis (like if Redis is the master) we need to support reading strings not crafted by Dragonfly which may use sparse encoding.
Or did I miss your intention? Sorry if so!

[romange]

You are right that it can happen during the rdb load or using the replication from Redis master.

1. Please add a comment about the case
2. Please add a test to rdb_test as well. This is how I did it. I run redis 6.2, added a couple of items , in this case pf counts, and then shutdown the server which produces dump.rdb. Then I rename it, move it to testdata and use it in rdb_test to show compatibility. We want both dense and sparse items be present in the rdb, I guess.

[chakaz]

I'm happy to add the rdb_test, but just making sure before: have you looked at the WorkWithSparseEncoding test that I added? I took that string from Redis, and so I'm not sure what additional coverage will rdb_test add?
I guess I could also add a Redis-created dense-encoding string, would that do?

[romange]

I guess rdb_test is not needed then. Yes, adding a dense string would suffice, thanks!

[chakaz]

Great, done!

[chakaz]

I switched to use rdb_test indeed, as explicitly writing the dense-encoded string was just too long (took > 1000 lines!).
In addition, I switched to use GetSlice() which should save a bunch of allocations, so nice optimization!

PTAL.

[romange]

Looks great!