perf(java): optimize map copy perf

[chaokunyang]

What does this PR do?

optimize map copy perf and add benchmark

Related issues

Closes #1744

Does this PR introduce any user-facing change?

• Does this PR introduce any public API change?
• Does this PR introduce any binary protocol compatibility change?

Benchmark

Benchmark                           (bufferType)  (references)   Mode  Cnt        Score         Error  Units
CopyBenchmark.fury_copy_int_map            array         false  thrpt    3  1779502.276 ± 2305262.034  ops/s
CopyBenchmark.fury_copy_string_map         array         false  thrpt    3  1845474.980 ± 2072340.777  ops/s
CopyBenchmark.kryo_copy_int_map            array         false  thrpt    3   625471.456 ±  289827.215  ops/s
CopyBenchmark.kryo_copy_string_map         array         false  thrpt    3   649970.300 ±  404700.716  ops/s

Benchmark                           (bufferType)  (references)   Mode  Cnt        Score         Error  Units
CopyBenchmark.fury_copy_int_map            array         false  thrpt    3  2023724.754 ± 2067096.856  ops/s
CopyBenchmark.fury_copy_string_map         array         false  thrpt    3  1979796.679 ±  593560.983  ops/s
CopyBenchmark.kryo_copy_int_map            array         false  thrpt    3   569146.293 ±  461667.597  ops/s
CopyBenchmark.kryo_copy_string_map         array         false  thrpt    3   591048.755 ±  669110.894  ops/s

[chaokunyang]

All remaining cost is map put, which we can't optimize