[BUG] ByteSizeCachingDirectory wrapped IndexOutput should delegate primitive write methods

[HUSTERGS]

Describe the bug

A Store will wrap the underlying Directory with ByteSizeCachingDirectory by default, and all the IndexOutput created from ByteSizeCachingDirectory will be wrapped again with FilterIndexOutput.

OpenSearch/server/src/main/java/org/opensearch/index/store/ByteSizeCachingDirectory.java

Lines 165 to 170 in 1c171b7

	return new FilterIndexOutput(out.toString(), out) {
	@Override
	public void writeBytes(byte[] b, int length) throws IOException {
	// Don't write to atomicXXX here since it might be called in
	// tight loops and memory barriers are costly
	super.writeBytes(b, length);

However, this wrap does not delegate primitive write methods like writeLong, writeInt, etc. , which makes optimization like apache/lucene#321 cannot work as expected, since the FilterIndexOutput does not delegate those methods too. So the those primitive write will eventually be written byte by byte (which is the default implementation of DataOutput).
By delegating those methods like what RateLimitedIndexOutput from Lucene does, the merge speed should be improved.

Please help me out if I got anything wrong : )

Related component

Indexing:Performance

To Reproduce

Check the code

Expected behavior

We should delegate those methods like what Lucene RateLimitedIndexOutput does, ref: https://github.com/apache/lucene/blob/main/lucene/core/src/java/org/apache/lucene/store/RateLimitedIndexOutput.java

Additional Details

Plugins
Please list all plugins currently enabled.

Screenshots
If applicable, add screenshots to help explain your problem.

Host/Environment (please complete the following information):

• OS: [e.g. iOS]
• Version [e.g. 22]

Additional context
Add any other context about the problem here.