fix: skip a UTF-8 BOM when reading config files

[geek-alpha]

Summary

git skips a UTF-8 BOM at the start of a config file. GitConfigParser raised MissingSectionHeaderError instead, because the decoded BOM became part of the first line, and that line then no longer matched a section header.

• Skip the BOM bytes when reading the first line of a config file.
• Add a regression test that fails without the change.

Why this is a bug and not malformed input

git reads such a file fine, so GitPython disagreeing with git here is the bug. The error is also hard to act on: the file clearly starts with [core], yet the message says it contains no section headers.

Validation

git 2.47.3, file starting with the three BOM bytes EF BB BF:

$ git config -f bom.cfg --list
core.bare=true
$ echo $?
0

GitPython before the change:

configparser.MissingSectionHeaderError: File contains no section headers.
file: 'bom.config', line: 1
'\ufeff[core]\n'

After the change get_value("core", "bare") is True, matching git. A BOM file reached through include.path is read the same as git as well (user.name=bom-inc).

Test runs:

• pytest test/test_config.py -o addopts= -q -> 48 passed, 2 skipped, 20 subtests passed
• pytest test/ -o addopts= -q -> 988 passed, 82 skipped, 1 xfailed
• ruff check / ruff format --check on both files -> clean

AI agent disclosure

Per CONTRIBUTING ("Prevent agent impersonation"): this PR was prepared by an AI
agent (Battle Phoenix) driving the geek-alpha account. The account owner reviewed
and approved the change, the test and the evidence above.

[Byron]

Great, thanks!