to start counting lines at 1 rather than 0, unless i've missed a +1 somewhere else!

you can re.compile this pattern once at the beginning and then match with that, rather than recompiling on each iteration of the loop

Why is that a benefit?

it might not be. in theory, compiling it once and then using it many times should be kinder on system resources. however, i believe python maintains its own cache for compiled regex patterns anyway, so you may see no difference at all.
i'm also not sure how long the average tmux config is; if this loop only runs a few hundred times it would be a negligible difference.
thought it was worth pointing out for consideration regardless.

edit: i do see some improvement locally:

$ python -m timeit -s 'import re' 're.match("[a-z]", "hello world!")'
1000000 loops, best of 5: 227 nsec per loop
$ python -m timeit -s 'import re; pat = re.compile("[a-z]")' 'pat.match("hello world!")'
5000000 loops, best of 5: 83.3 nsec per loop

It is cached, and I'm not really interested in micro optimisations without a measurable performance benefit that users would notice

as i said, just a suggestion :)