gh-117716: Fix wave RIFF padding for data chunks

[mbeijen]

wave.Wave_write now writes the required RIFF pad byte when the data chunk size is odd.

Update RIFF chunk size calculations in both header writing and header patching so they include the alignment pad byte when present.

Add a regression test in test_wave.py that verifies odd-sized writes are padded, RIFF size is correct, and roundtrip reads preserve frame data.

• Issue: Wave_write does not write pad byte #117716

[encukou]

Looking at the document referenced in the issue (emphasis mine):

Binary data of fixed or variable size. The start of ckData is word-aligned with respect to the start of the RIFF file. If the chunk size is an odd number of bytes, a pad byte with value zero is written after ckData. Word aligning improves access speed (for chunks resident in memory) and maintains compatibility with EA IFF. The ckSize value does not include the pad byte.

In this PR, the size includes the padding byte. Which is right?

[encukou]

@mbeijen, do you have an answer?

[mbeijen]

@mbeijen, do you have an answer?

Sure, and thanks for following up!! Plus, I've rebased the commit so it now applies cleanly after my other PRs affecting wave.py plus the tests.

The spec quote "ckSize does not include the pad byte" applies to each sub-chunk's own ckSize. The RIFF form's ckSize represents total file content minus the 8-byte RIFF header, which physically includes the pad byte. I double-checked this with libsndfile (what audacity uses) and it does exactly the same thing as this fix.

I verified this against libsndfile 1.2.2: for 1 byte of audio it writes data ckSize = 1 and RIFF ckSize = 38 (= 36 + 1 + 1), matching this PR exactly.

[encukou]

I see, that makes sense. Thanks for double-checking!