chore: Batch aggregator & range aggregation pipeline

[benclive]

What this PR does / why we need it:
Switches the aggregator to operate on batches in order to be more performant.

I achieved a 2-3x improvement to range aggregation using this approach (+ operating on UnixNano instead of time.Time was significant)

RangeAggregation:

goos: darwin
goarch: arm64
pkg: github.com/grafana/loki/v3/pkg/engine/internal/executor
cpu: Apple M3 Max
                                             │ range_agg_before.txt │        range_agg_after.txt         │
                                             │        sec/op        │   sec/op     vs base               │
RangeAggregationPipeline/case=instant-14               189.17µ ± 1%   84.50µ ± 2%  -55.33% (p=0.002 n=6)
RangeAggregationPipeline/case=aligned-14                82.36µ ± 6%   34.54µ ± 3%  -58.06% (p=0.002 n=6)
RangeAggregationPipeline/case=gapped-14                 80.97µ ± 2%   30.97µ ± 1%  -61.75% (p=0.002 n=6)
RangeAggregationPipeline/case=overlapping-14            86.98µ ± 3%   38.96µ ± 2%  -55.21% (p=0.002 n=6)
geomean                                                 102.3µ        43.32µ       -57.68%

                                             │ range_agg_before.txt │        range_agg_after.txt         │
                                             │         B/op         │     B/op      vs base              │
RangeAggregationPipeline/case=instant-14               21.92Ki ± 0%   20.64Ki ± 0%  -5.81% (p=0.002 n=6)
RangeAggregationPipeline/case=aligned-14               23.14Ki ± 0%   21.86Ki ± 0%  -5.51% (p=0.002 n=6)
RangeAggregationPipeline/case=gapped-14                21.18Ki ± 0%   20.03Ki ± 0%  -5.43% (p=0.002 n=6)
RangeAggregationPipeline/case=overlapping-14           24.26Ki ± 0%   25.38Ki ± 0%  +4.61% (p=0.002 n=6)
geomean                                                22.59Ki        21.89Ki       -3.13%

                                             │ range_agg_before.txt │        range_agg_after.txt        │
                                             │      allocs/op       │ allocs/op   vs base               │
RangeAggregationPipeline/case=instant-14                 274.0 ± 0%   234.0 ± 0%  -14.60% (p=0.002 n=6)
RangeAggregationPipeline/case=aligned-14                 296.0 ± 0%   256.0 ± 0%  -13.51% (p=0.002 n=6)
RangeAggregationPipeline/case=gapped-14                  265.0 ± 0%   234.0 ± 0%  -11.70% (p=0.002 n=6)
RangeAggregationPipeline/case=overlapping-14             301.0 ± 0%   278.0 ± 0%   -7.64% (p=0.002 n=6)
geomean                                                  283.6        249.8       -11.90%

Aggregator itself:

goos: darwin
goarch: arm64
pkg: github.com/grafana/loki/v3/pkg/engine/internal/executor
cpu: Apple M3 Max
                                    │ agg_before.txt │           agg_after.txt            │
                                    │     sec/op     │   sec/op     vs base               │
AggregatorAdd/case=interleaved-14       119.57µ ± 1%   69.39µ ± 1%  -41.97% (p=0.002 n=6)
AggregatorAdd/case=single_series-14     118.47µ ± 1%   69.01µ ± 2%  -41.75% (p=0.002 n=6)
geomean                                  119.0µ        69.20µ       -41.86%

                                    │ agg_before.txt │           agg_after.txt           │
                                    │      B/op      │    B/op     vs base               │
AggregatorAdd/case=interleaved-14        382.00 ± 1%   26.00 ± 0%  -93.19% (p=0.002 n=6)
AggregatorAdd/case=single_series-14      373.00 ± 0%   21.50 ± 2%  -94.24% (p=0.002 n=6)
geomean                                   377.5        23.64       -93.74%

                                    │ agg_before.txt │             agg_after.txt              │
                                    │   allocs/op    │ allocs/op   vs base                    │
AggregatorAdd/case=interleaved-14         2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.002 n=6)
AggregatorAdd/case=single_series-14       2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.002 n=6)
geomean                                   2.000                    ?                      ¹ ²
¹ summaries must be >0 to compute geomean
² ratios must be >0 to compute geomean

Note: I used AI to make most of the changes and committed after checking each major change.

Best commits to review:
Support without
Use unixnano everywhere instead of time.Time
and the final Clean up