Fix fetching licenses for jvm packages

[puchta]

I've found that license information of mvn packages may be stored in at least 4 ways (so far cdxgen used only the fist one).

• in the field licenses.license of the POM document
• in the field licenses.license of another POM document (defined by the parent field of the POM).
• as a comment in the POM.xml
• in the github repo specified in the POM document

I've implemented

• merging POM witht the parent POM
• regexping pom in search of license text
• querying the repo (if specified in POM)
  • I've improved the way github api address is deduced from repo address - so that now it accepts more patterns.

[puchta]

@prabhu what do you think?

[prabhu]

This looks so good! Is it possible to add a test to repo tests?

[prabhu]

cc: @heubeck

[puchta]

Is it possible to add a test to repo tests?

OK, I'm working on it.

[prabhu]

@puchta, would you like to rebase and continue with the repo test?

[prabhu]

@puchta, ready to merge this PR and release 9.9.1. If you could contribute a test in a new PR, that would be awesome.

[theold190]

Minor typo: Publich

[prabhu]

Could you send a new PR? It is possible that the typo is deliberate so that it matches with the correct spdx id but let's not implement stemming algorithms in cdxgen (or not yet ;))

[puchta]

The typeo is deliberate, see https://central.sonatype.com/artifact/org.scalanlp/breeze_2.12/1.3?smo=true

[puchta]

would you like to rebase and continue with the repo test?

Thanks for merging.
Yes, I'll push the tests as soon as I have them.