fix: Reader proceeds to next volume when current one runs out of bytes

[jaens]

Currently, extracting IS5 CABs that split files over multiple volumes seems to fail with bytes_to_read can't be zero, because the current volume runs out of bytes, meaning MIN(bytes_left, reader->volume_bytes_left) becomes 0 which always leads to an error in unshield_reader_read.

Detect this condition and jump to opening the next volume. This is possibly related to #124 and #129, although it might not fix the latter.

[twogood]

goto is not something you see everyday! :)

Do you have some .cab files that can be used as test cases?

[twogood]

I wonder why the tests fail :(

[jaens]

goto is not something you see everyday! :)

Can be avoided by restructuring all the surrounding code into nested ifs, I suppose...

Do you have some .cab files that can be used as test cases?

uhh... not sure if abandonware can be distributed as a test case?

I wonder why the tests fail :(

I'll check them locally, might need even more extra checks to prevent infinite loops, in case something else runs out I guess.

[twogood]

@jaens The existing test cases are pretty much based on abandonware... I need to check that they have not started to fail on main branch as well.

[twogood]

All tests succeeded in #174 so I can't take in this PR in its current state.

[twogood]

• Existing tests must still work
• Add new test case for this issue (just cut & paste an existing test)

(The goto can be refactored later.)