base64: don't drop already-decoded output when decoding hits invalid input

[Chaganti-Reddy]

base64 -d was throwing away everything it had already decoded whenever it hit invalid input later in the stream, instead of writing out the good prefix like GNU does. root cause was two spots that only flushed whole internal batches instead of whatever complete groups were actually buffered, plus base64's decoder rejecting a trailing group with non-canonical padding bits that GNU still accepts.

Fixes #12923

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/pr/bounded-memory (fails in this run but passes in the 'main' branch)
Skip an intermittent issue tests/tail/tail-n0f (fails in this run but passes in the 'main' branch)
Congrats! The gnu test tests/expand/bounded-memory is now passing!
Congrats! The gnu test tests/rm/many-dir-entries-vs-OOM is now passing!

[codspeed]

Merging this PR will degrade performance by 32.13%

⚡ 2 improved benchmarks
❌ 1 regressed benchmark
✅ 394 untouched benchmarks
🆕 2 new benchmarks
⏩ 54 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	three_39_bit_primes	228.6 ms	797.8 ms	-71.35%
⚡	Simulation	du_wide_tree[(5000, 500)]	15.9 ms	15.2 ms	+4.56%
⚡	Simulation	tr_single_char_replace	14.5 ms	13.9 ms	+4.36%
🆕	Memory	tr_delete_single_char	N/A	35.4 KB	N/A
🆕	Simulation	tr_delete_single_char	N/A	103.9 ms	N/A

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing Chaganti-Reddy:base64-partial-decode-on-error (3c6673a) with main (a185142)²}

Footnotes

1. 54 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. No successful run was found on main (9768800) during the generation of this report, so a185142 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩