Partitions.db internally uses a piece of the partition key murmur
hash (the same hash which is used to compute the token and the
relevant bits in the bloom filter). Before this patch,
the Partitions.db reader computes the hash internally from the
`sstables::partition_key`.
That's a waste, because this hash is usually also computed
for bloom filter purposes just before that.
So in this patch we let the caller pass that hash instead.
The old index interface, without the hash, is kept for convenience.
In this patch we only add a new interface, we don't switch the callers
to it yet. That will happen in the next commit.
Before this patch, `bti_index_reader::last_block_offset()` returns the
offset of the last block within the file.
But the old `index_reader::last_block_offset()` returns the offset
within the partition, and that's what the callers (i.e. reversed
sstable reader) expect.
Fix `bti_index_reader::last_block_offset()` (and the corresponding
comment and test) to match `index_reader::last_block_offset()`.
Implements an index reader (implementing abstract_index_reader,
which is the interface between Data.db readers and index readers)
for the BTI index written by bti_partition_index_writer and bti_row_index_writer.
The underlying trie walking logic is a thin wrapper around the logic
added in `trie_traversal.hh` in an earlier patch series.
In later parts of the series, we add a second implementation
of `abstract_index_reader`. To do that, we want a header with
`abstract_index_reader`, but we don't need to pull in everything else
from `index_reader.hh`. So let's extract `abstract_index_reader`
to its own header.
