Improve Handling of Cutoff Dates via CLI

[johannesduesing]

As reported in #48, we were missing proper documentation for the usage of the time-based artifact filtering via --since-until. For MARIN 1.0.0, we require both cutoff dates to be specified as UNIX timestamps in milliseconds. While this is what Maven Central provides, it is neither a common format nor very intiutive to users. Therefore, this PR:

• Changes the accepted format for cutoff dates provided via the --since-until CLI argument. Dates can now be given as either UNIX timestamps in seconds, or as dates in the YYYY-MM-DD format. For the latter format, the beginning of a time range (i.e., the since value) will be interpreted as the beginning of the specified date (time 00:00), while the range end date (i.e., the until value) will be interpreted as end of day (time 23:59). Specifying --since-until 2025-12-01:2025-12-02 will thus include all artifacts release on either December 1st or December 2nd, 2025.
• Updates README to actually include an example usage of --since-until showcasing the new formats.
• Updates ArtifactAnalysisConfigBuilder to include a convenience method that accepts a ZonedDateTime instead of plain long values for cutoff dates.

Note that YYYY-MM-DD dates are converted into timestamps based on the user's default time zone Dates are converted into timestamps using the GMT timezone - to fine-tune cutoff dates, UNIX time stamps should be used.

Closes #48.

[jachiaram]

The changes look good, though the test is failing due to no maven-metadata file being found for the library analysis. For a library to be analyzed, does it need the maven-metadata file to get all versions?

[jachiaram]

Or a better question is for the given skip take configuration, if a library can't be processed should it still be counted as a part of the take?

[johannesduesing]

It took me some time, but i'm finally coming back to this PR 😅 I did a few things to address test failures:

• Used the onVersionListError hook in tests to also count libraries where metadata.xml does not exist - it's to be expected for older libraries, and is not an error condition, so in my mind it should count as part of the "take"
• It seems the Maven Central Index has been altered - At least the first few hundert GAV triples are not the same anymore compared to when we wrote some of our tests. This means we cannot realy on the index order for our unit tests - some of them were failing because of this exact issue. I changed all test that relied on index order (some specific GAV being at some specific position in the index) to use an input list of explicit GAVs that we have control over.

This should fix our test issues. I'm just now seeing a build failure, but its due to some GitHub Internal Server Error when uploading the dependency tree and should be unrelated - i'll do some research and re-trigger the CI run.