MINOR: Close already-opened input readers when ParquetRewriter setup fails

[anxkhn]

Rationale for this change

ParquetRewriter.getFileReaders opens a TransParquetFileReader for each input
file into a local list. If opening a later file throws IOException, it is
rethrown as IllegalArgumentException without closing the readers already opened
for the earlier files, so their SeekableInputStream handles leak.
ParquetRewriter.close() only ends the writer and does not close the input
readers, so nothing else releases them either. A single corrupt, missing, or
permission-denied file among several valid inputs therefore leaks up to N open
input streams per failed rewrite/merge/join setup.

What changes are included in this PR?

In getFileReaders, close the readers already opened before rethrowing, using
org.apache.parquet.util.AutoCloseables.close(...) so every reader is released
and any close failures are aggregated as suppressed exceptions on the original
IllegalArgumentException. This matches the existing cleanup idiom in the module.
The change is limited to the catch (IOException) block; the success path is
unchanged.

Are these changes tested?

Yes. A new ParquetRewriterTest case
(testInputFileReadersClosedWhenLaterInputFileFailsToOpen) supplies a valid
input file (wrapped so it records whether the stream it hands out is closed)
followed by an InputFile whose newStream() throws, asserts that constructing
the ParquetRewriter throws IllegalArgumentException, and asserts the first
reader's stream was closed. It fails on the pre-fix code (the first stream is
left open) and passes with the fix. The full ParquetRewriterTest class passes
(Tests run: 96, Failures: 0, Errors: 0).

Are there any user-facing changes?

No public API changes. The only behavioral difference is that a failed
ParquetRewriter setup no longer leaks the input streams it had already opened.

[Fokko]

Seams reasonable to me 👍

[wgtmac]

I would recommend creating an issue for this. Or at least do not mark it as MINOR.