It's not so much about the file extension, but

1. how the data is collected
2. how the data is formatted
3. how the line order is determined

I can see 2 simple options that might help at least with
parts of the request.

a) txt

That might be the actual output from user-space echo()
calls only, both function and module.
No guarantee about ordering of lines to be stable across
future OpenSCAD releases!

b) csv

That could use the naming feature of echo(), e.g.
using echo(a = 1, b = "hello"); would generate 2 columns
named "a" and "b". Unnamed values may be dropped or may
be added in some other defined way. Again no ordering
guarantee.

As for BOM output, I'm still pointing to the meta data
use cases as I think that's an even better solution and
in-line with the descriptive nature of OpenSCAD:
https://github.com/openscad/openscad/wiki/Meta-Data-Use-Cases

ciao,
Torsten.

Any sort of file output which was directly usable and didn't force the need for post processing (as preceding each line w/ "ECHO: " does now) would help.
The mata data for B.O.M. is interesting --- are there any implementations yet? Looks like it could be used for a lot of other things as well, possible even CAM.
The ordering issue is an interesting one --- G-Code apparently can use line #s, so the responsibility for this could be pushed downstream (program adds line #s based on intended file processing, user checks file and sorts it if need be).
William