The previously mentioned issue with changing time signatures (bug ID 4658) will be fixed in an upcoming build that I'm currently testing.

As well, thanks to Gordon's file, I've identified a number of issues:

Originally Posted by Gordon Scott, via email
With the default two-bar offset BiaB tries to do cut-time in quarter-notes, not half notes. With a more suitable 1-bar offset it does 4/4 (OK) but has the anacrucis left justified.
Reported with bug id 4664 and 4665.

Originally Posted by Gordon Scott, via email
It fails to recognise several Majors.
Will be fixed in an upcoming build (bug id 2916).

Originally Posted by Gordon Scott, via email
It sometimes drops the first chord in bars. (IMHO It fails where there are too many chords per bar, but I guess that's life).
Reported with bug id 4667.

Originally Posted by Gordon Scott, via email
It may not recognise the Signe on bar one. It doesn't recognise first and second ending (maybe that's an expected limitation). It doesn't appear to recognise the D.S. al Coda (ditto). I'm not sure whether it recognises/handles the Coda.
Reported with bug id 4666.

Originally Posted by Gordon Scott
I must also say that opening/importing MusicXML open things up to all sorts of snags in all sorts of places, not necessarily BIAB, though I tried opening it with Sibelius and it seemed very close to the original.
That certainly is the case - as much as MusicXML is supposed to be a universal format, every program handles things differently. This makes it incredibly hard to achieve accurate and predictable results (for all software, not just BIAB).

Originally Posted by Gordon Scott
The combination of my very gradually improving depression issues and the frustration do now sometimes make it difficult for me to do these tests.
I know the feeling. I hope things improve for you.

I work here