Core: Use zero-copy wrapper for equalityFieldIds

[bvolpato]

In one of our Trino -> Iceberg use case, which relies on equality deletes, we observed that a lot of time and allocations are being spent in BaseFile.equalityFieldIds() (~34% of our overall allocs).

Checking the implementation, it seems that the current implementation based on streams has to copy and box the data, which is very inefficient.

There is already prior art (#8336) to use Guava 0-copy wrappers for longs / split offsets in the same class, I'm doing that same change.

Drafted a quick JMH to show the difference, and it clearly makes a huge difference:

Benchmark                              (arraySize)  Mode  Cnt        Score         Error  Units
IntListBenchmark.guavaImplementation            10    ss   10     1266.800 ±     199.548  ns/op
IntListBenchmark.guavaImplementation          1000    ss   10     1387.200 ±     531.909  ns/op
IntListBenchmark.guavaImplementation        100000    ss   10     1262.500 ±     188.043  ns/op
IntListBenchmark.guavaImplementation       1000000    ss   10     1600.000 ±    1509.810  ns/op
IntListBenchmark.streamImplementation           10    ss   10     8883.500 ±    1858.339  ns/op
IntListBenchmark.streamImplementation         1000    ss   10    46229.000 ±   25265.952  ns/op
IntListBenchmark.streamImplementation       100000    ss   10   736904.100 ±  299912.894  ns/op
IntListBenchmark.streamImplementation      1000000    ss   10  7321966.800 ± 7146920.166  ns/op

[amogh-jahagirdar]

This is a great find @bvolpato, the improvement makes a lot of sense to me!

[amogh-jahagirdar]

Thanks for the PR @bvolpato and thank you @mrcnc @pvary for reviewing. I'll go ahead and merge

[findinpath]

Could you pls share the code of IntListBenchmark ?

[findinpath]

It is worth considering rewriting org.apache.iceberg.util.ArrayUtil#toIntList as well even though it is used at the moment only in tests.

[rdblue]

Just found this and I agree.

[wendigo]

@bvolpato off topic: what’s the app from the screenshot?

[bvolpato]

Could you pls share the code of IntListBenchmark ?

@findinpath Sorry, it was a bit of throwaway code, but I still had it saved here, pushed to https://github.com/bvolpato/guava-temp-benchmark/blob/main/src/jmh/java/com/benchmark/IntListBenchmark.java

@bvolpato off topic: what’s the app from the screenshot?

@wendigo (Disclaimer: I work at Datadog) It's from Datadog continuous profiling https://docs.datadoghq.com/profiler/, it has awesome instrumentation/tooling to proactively sample applications with very little overhead, which allows us to just go back in time for a particular time that had CPU/memory pressure and look at the flamegraphs - which was the case here.