chore: Add alternative int64 encoder to dataobj

[benclive]

What this PR does / why we need it:
Adds an alternative implementation encoding for int64s in dataobjects.

• Inspiration from https://lemire.me/blog/2012/09/12/fast-integer-compression-decoding-billions-of-integers-per-second/ and the referenced Go implementation https://github.com/ronanh/intcomp
• I didn't achieve the stated speeds of 14-24GB/s on int64s but it still turned out to be better on all fronts compared to regular delta encoding.
  • Faster encoding (25% faster sequential, 2.5x faster random)
  • Faster decoding (2.7x faster sequential, 7.9x faster random (!))
  • Smaller file sizes (0.25-8.133 bytes/value vs 1-10 bytes per value using delta)

I think the faster decoding speed is really beneficial to query perf, the other improvements are just bonuses. I haven't hooked this up in any builders yet, but I can do that afterwards if we want to use this.

Raw benchmark results:

goos: darwin
goarch: arm64
pkg: github.com/grafana/loki/v3/pkg/dataobj/internal/dataset
cpu: Apple M3 Max
Benchmark_intCompEncoder_Encode/Sequential-14         	277998073	         4.297 ns/op	         1.862 GB/s	       0 B/op	       0 allocs/op
Benchmark_intCompEncoder_Encode/Largest_delta-14      	233474617	         5.122 ns/op	         1.562 GB/s	       0 B/op	       0 allocs/op
Benchmark_intCompEncoder_Encode/Random-14             	192020827	         6.072 ns/op	         1.317 GB/s	       0 B/op	       0 allocs/op
Benchmark_deltaEncoder_Encode/Sequential-14         	222049738	         5.392 ns/op	         1.484 GB/s	       0 B/op	       0 allocs/op
Benchmark_deltaEncoder_Encode/Largest_delta-14      	87491991	        13.73 ns/op	         0.5825 GB/s	       0 B/op	       0 allocs/op
Benchmark_deltaEncoder_Encode/Random-14             	77057061	        15.66 ns/op	         0.5108 GB/s	       0 B/op	       0 allocs/op
PASS
ok  	github.com/grafana/loki/v3/pkg/dataobj/internal/dataset	4.382s

goos: darwin
goarch: arm64
pkg: github.com/grafana/loki/v3/pkg/dataobj/internal/dataset
cpu: Apple M3 Max
Benchmark_intCompDecoder_Decode/Sequential-14         	705083431	         1.682 ns/op	         4.758 GB/s	         0.2539 bytes/value	       0 B/op	       0 allocs/op
Benchmark_intCompDecoder_Decode/Largest_delta-14      	510498273	         2.337 ns/op	         3.423 GB/s	         8.133 bytes/value	       0 B/op	       0 allocs/op
Benchmark_intCompDecoder_Decode/Random-14             	516526146	         2.392 ns/op	         3.344 GB/s	         8.133 bytes/value	       0 B/op	       0 allocs/op
Benchmark_deltaDecoder_Decode/Sequential-14         	259016820	         4.577 ns/op	         1.748 GB/s	         1.000 bytes/value	       0 B/op	       0 allocs/op
Benchmark_deltaDecoder_Decode/Largest_delta-14      	66355470	        18.16 ns/op	         0.4405 GB/s	        10.00 bytes/value	       0 B/op	       0 allocs/op
Benchmark_deltaDecoder_Decode/Random-14             	64460536	        18.93 ns/op	         0.4226 GB/s	         9.242 bytes/value	       0 B/op	       0 allocs/op
PASS
ok  	github.com/grafana/loki/v3/pkg/dataobj/internal/dataset	9.529s