scanner: add support for an exclusion list

When scanning local code there are often file patterns that we want to exclude as it doesn't make sense to scan them. Common offenders will be things like .git directories, .tox, but also node_modules/ and a bunch of others, depending on the case.

Hence we want to have:

• the ability to specify exclusion patterns on the command line, e.g., with a -x/--exclude option; it should be possible to pass it multiple times
• the ability to specify exclusion patters in the configuration file of swh-scanner
• some sensible defaults

The syntax and semantics of exclusion patterns is to be defined. Glob patterns might be an option. We need something that is expressive, common, and that we do not need to implement by hand.

---

Migrated from T2336 (view on Phabricator)