Fuzz Introspector heuristics suggest the Submodule API code represent
"optimal analysis targets" that should yield a meaningful increase in
code coverage. The changes here introduce a first pass at implementing a
fuzz harness that cover the primary APIs/methods related to Submodules.
Of particular interest to me is the `Submodule.config_writer()`
coverage.

Please note however, there is likely plenty of room for improvement in
this harness in terms of both code coverage as well as performance; the
latter of which will see significant benefit from a well curated seed
corpus of `.gitmodules` file like inputs. The `ParsingError` raised by
the fuzzer without a good seed corpus hinders test efficacy
significantly.