chunk: reduce chunk's iterator function call

[crazycs520]

What problem does this PR solve?

Reduce chunk's iterator function call.

What is changed and how it works?

Add numRows int to Iterator4Chunk.

Below is a benchmark

func BenchmarkIterator4Chunk_Next(b *testing.B) {
        // 1024 rows
	chk := getChks()
	it := NewIterator4Chunk(chk)
	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		for row := it.Begin(); row != it.End(); row = it.Next() {
			_ = row
		}
	}
}

old:

BenchmarkIterator4Chunk_Next     2000000              2712 ns/op
BenchmarkIterator4Chunk_Next     2000000              2711 ns/op
BenchmarkIterator4Chunk_Next     2000000              2704 ns/op

this branch

BenchmarkIterator4Chunk_Next     3000000              1412 ns/op
BenchmarkIterator4Chunk_Next     3000000              1411 ns/op
BenchmarkIterator4Chunk_Next     3000000              1410 ns/op

[eurekaka]

LGTM

[eurekaka]

Is there any use case that we are iterating the chunk while appending into it simultaneously? I guess this is not allowed, just to make sure.

[crazycs520]

@zz-jason

[zz-jason]

No, there isn't.

[zz-jason]

No, there isn't.

[zz-jason]

Can we change the type from int to int32 to make Iterator4Chunk only take 16 bytes in a 64-bit machine?

[zz-jason]

LGTM

[crazycs520]

/run-all-tests