[SPARK-54571][CORE] Use LZ4 safeDecompressor

[Yicong-Huang]

What changes were proposed in this pull request?

Switch LZ4CompressionCodec.compressedInputStream to use lz4Factory.safeDecompressor() instead of lz4Factory.fastDecompressor().

Since Spark upgraded to at.yawk.lz4:lz4-java:1.10.0 (via #53327), LZ4BlockInputStream now has a Builder API that accepts LZ4SafeDecompressor. This PR uses that API:

LZ4BlockInputStream.newBuilder()
  .withDecompressor(lz4Factory.safeDecompressor())
  .withChecksum(xxHashFactory.newStreamingHash32(defaultSeed).asChecksum)
  .withStopOnEmptyBlock(false)
  .build(s)

This is a follow-up to #53290, which was blocked because LZ4BlockInputStream in lz4-java 1.8.0 did not accept LZ4SafeDecompressor. The lz4-java 1.10.0 Builder API unblocks this change.

Why are the changes needed?

As noted by the lz4-java maintainer (@yawkat) in #53290, as of lz4-java 1.8.1, safeDecompressor is substantially faster than fastDecompressor. The recommended migration path with proper performance is to use safeDecompressor via the 1.10.0+ Builder API.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Existing unit tests.

Was this patch authored or co-authored using generative AI tooling?

No.

[dbtsai]

Do you have benchmark? Thanks.

[pan3793]

You guys might have missed #53454

[Yicong-Huang]

You guys might have missed #53454

Oops, yeah sorry we have missed your PR. This was brought into sight again due to performance regression. I will close this one and let's move our discussion in your PR.