Mimir query engine: improve performance for case where right side of binary operation has fewer series than the left side

[charleskorn]

What this PR does

This PR improves the performance of binary operations for queries where the right side has fewer series than the left side.

Benchmark results for a query where the left side (a_2000) has 2000 series, and the right side has one series:

                                                                                    │ original.txt  │             after.txt              │
                                                                                    │    sec/op     │   sec/op     vs base               │
Query/a_2000_-_b_2000{l="1234"},_instant_query/streaming-10                            6.917m ±  4%   6.327m ± 1%   -8.53% (p=0.002 n=6)
Query/a_2000_-_b_2000{l="1234"},_range_query_with_100_steps/streaming-10              10.391m ±  6%   9.670m ± 1%   -6.94% (p=0.002 n=6)
Query/a_2000_-_b_2000{l="1234"},_range_query_with_1000_steps/streaming-10              47.98m ±  4%   46.55m ± 1%   -2.98% (p=0.041 n=6)

Latency and peak memory utilisation is unchanged for other benchmarks, and peak memory utilisation is only marginally affected for the benchmark above.

Which issue(s) this PR fixes or relates to

Resolves a TODO from #8096

Checklist

• Tests updated.
• [n/a] Documentation added.
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX].
• [n/a] about-versioning.md updated with experimental features.

[jhesketh]

Nice, lgtm!