Updates stephan

[stschiff]

I am working on a few updates and some refactoring of #264

[nevrome]

Excellent - some last comments (I hope):

• I think list should have an extra, boolean column for the latest property. A (latest) suffix to the version number is harder to parse, especially for quick analyses on the command line.
• A number of subcommands (genoconvert, rectify, survey, summarize, validate) now only operate on the latest package versions in a given base directory. This may be confusing -- especially if there is no clear need to only consider one iteration of a package (e.g. in genoconvert, survey or validate). Should we make them multi-version compatible? Maybe they should behave like list: Always consider all package versions, except the --onlyLatest flag is set. This doesn't fully match what users might expect for rectify and summarize, but it would be a universal and consistent solution. We could essentially set _readOptKeepMultipleVersions in defaultPackageReadOptions to true.
• The technical, internal changelog should probably also mention the high-level refactoring within the library: The SecondaryTypes module was dissolved and instead we now have Contributor, ServerClient and Version. EntitiesList (or what is left of it) now lives in EntityTypes.
• The way we handle versions now (e.g. in parseVersion) is a bit imprecise. According to the implementation we only accept versions of the form X.Y.Z, but the Poseidon v2.7.1 standard does not specifically say this. I think we should make this constraint explicit in the next version of the standard.

Let me know if you agree with these ideas and whether I should take over and implement some of them.

[stschiff]

OK, list now has its own column for "Is latest". I also noticed that the raw output so far did not print out the column headings, which is however vital, in particular with additional janno columns when listing individuals, so I now added this to the raw output (see for example golden test output for test-pipeline list4.

[stschiff]

Regarding the second point, I have now changed the default behaviour. I looked at the golden test outputs, and think that summarise looks strange when applied to multiple versions. I think it makes no sense in that specific case, so I explicitly now set _readOptKeepMultipleVersions = False in summarise. Happy to discuss

[stschiff]

Regarding points 3 and 4 I agree! I think you wanted to take a stab at the release changelog, and regarding point 4 yes, I will make a PR to the schema (no Poseidon version change, just a clarification, I think)

[stschiff]

OK very cool, I've reviewed the code changes and ran a few command line tests on the test-data. I'm happy. What do you say? I say, we merge and start a test-release, then get the rest of the ball going. Of course, finalist the release only after further tests and update of the server.

[stschiff]

I've added a fix for validate, which would fail if some packages were skipped if --onlyLatest is selected.