aac: refactor approximate frame count, fuzzy fixes

[sscobici]

refactored aac aproximate_frame_count() method. It had some issues which causes panic. Should fix cases from https://github.com/qarmin/Automated-Fuzzer/actions/runs/12516290494 (REPORTS___SYMPHONIA*)

1. renamed total_len to remaining_len because this is not the length of the whole file, but it was considered such in some places.
2. first sample point to start from the beginning, file can contain a single frame
3. check step to be bigger than 0 to avoid panic
4. skip the rest of the frame after the header is read to avoid looking for sync word in the audio data
5. break if reading 100 frames overlap next sample point

I haven't tested on VBR aac to observe any improvements, It was #196 (comment) approximate method is not perfect there

[pdeljanov]

Nice catch for an optimization! However, I think ignore_bytes will work even better since the MSS will decide if it should just consume from it's internal ring buffer or resort to an actual seek. In other words, you can just ignore_bytes(header.frame_len).

[pdeljanov]

A slightly more descriptive name.

Suggested change

	const NUM_FRAMES: u32 = 100;
	const NUM_FRAMES_PER_SAMPLE: u32 = 100;

[sscobici]

Thanks for suggestion.
ignore_bytes should ignore only the frame.payload because sync word and header were already read.
added another exit point in a loop to break if searching within the sample point overflow the next sample point. Can happen when no sync word found in a very broken file.