GH-48864: [C++] Support customizing more Zstd parameters

[HuaHuaY]

Rationale for this change

Support customizing more Zstd parameters to allow users to optimize the performance of parquet page compression/decompression.

What changes are included in this PR?

1. Add a new class ZstdCodecOptions inherited from CodecOptions.
2. Create ZSTD_CCtx/ZSTD_DCtx explicitly in ZSTDCodec.

Are these changes tested?

Yes.

Are there any user-facing changes?

Yes. The new class ZstdCodecOptions is added.

• GitHub Issue: [C++] Support customizing more Zstd parameters #48864

[github-actions]

⚠️ GitHub issue #48864 has been automatically assigned in GitHub to PR creator.

[HuaHuaY]

@pitrou @wgtmac @mapleFU Please take a look. This PR only uses context and does not reuse it. I create a new context every time I call Compress and Decompress methods. I guess this is the same behavior as before.

[pitrou]

Thank you @HuaHuaY . This can be a useful improvement, I've made a bunch of comments below.

[HuaHuaY]

@pitrou Thank you for your detailed review! I've fixed these comments. Please take a look again.

[pitrou]

This looks mostly good! Just a few nits.

[pitrou]

Neat, a concept 😀

[pitrou]

Funny that the compression level is passed a second time here, after we already gave it to ZSTD_CCtx_setParameter. I suppose it doesn't matter?

[HuaHuaY]

Context will also be passed to ZSTDCompressor and it needs compression_level_. We can move auto ret = ZSTD_CCtx_setParameter(cctx.get(), ZSTD_c_compressionLevel, compression_level_) to MakeCompressor though I think the readability will decrease a bit.

[HuaHuaY]

After I check the Zstd's document again, I think there is a mistake here. We need to use ZSTD_compress2 instead of ZSTD_compressCCtx because ZSTD_compressCCtx will reset other parameters. Thank you for your good attention!

/*! ZSTD_compressCCtx() :
 *  Same as ZSTD_compress(), using an explicit ZSTD_CCtx.
 *  Important : in order to mirror `ZSTD_compress()` behavior,
 *  this function compresses at the requested compression level,
 *  __ignoring any other advanced parameter__ .
 *  If any advanced parameter was set using the advanced API,
 *  they will all be reset. Only @compressionLevel remains.
 */
ZSTDLIB_API size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
                                     void* dst, size_t dstCapacity,
                               const void* src, size_t srcSize,
                                     int compressionLevel);

/*! ZSTD_compress2() :
 *  Behave the same as ZSTD_compressCCtx(), but compression parameters are set using the advanced API.
 *  (note that this entry point doesn't even expose a compression level parameter).
 *  ZSTD_compress2() always starts a new frame.
 *  Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
 *  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
 *  - The function is always blocking, returns when compression is completed.
 *  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
 *        enough space to successfully compress the data, though it is possible it fails for other reasons.
 * @return : compressed size written into `dst` (<= `dstCapacity),
 *           or an error code if it fails (which can be tested using ZSTD_isError()).
 */
ZSTDLIB_API size_t ZSTD_compress2( ZSTD_CCtx* cctx,
                                   void* dst, size_t dstCapacity,
                             const void* src, size_t srcSize);

[HuaHuaY]

I have added a new test TEST(TestCodecMisc, ZstdLargerWindowLog) to ensure the other parameters work effectively.