Keep at least one result for min-transfers and each transit-group in itinerary-group-filter

[t2gran]

Summary

The MaxLimit filter in GroupByFilter in the ItineraryListFilterChain did only look at generalized-cost when it reduced the number of itineraries in each group. This problematic because the itinerary with the best number of transfers or good alternatives for transit-groups could be lost. The new McMaxLimit filter will look at all pareto criterias used in the Raptor search and keep at least one optimal result for each criteria, before limiting the result to N itineraries.

To enable this the OTPFeature.MultiCriteriaGroupMaxFilter must be set in otp-config.json.

Issue

There is no issue for this, but one case we have seen at Entur is:

The alternative with two legs with the same agency is filtered away, in favor of a slightly better cost alternative with two legs with different agencies. The first leg is the same, but because of the switch to another agency the customer has to bye two tickets and have no guarantee in case of delays. So we want both options to be shown.

This is related to #5436 and #3833.

Unit tests

✅ All new business logic has unit-tests with 100% coverage.

Documentation

✅ There is extensive JavaDoc documentation, and a little bit of explaination on the new OTPFeature to enable this behaivour. OTP will work as before, if the feature is not enabled in config.

Changelog

✅ This is a minor improvment

Bumping the serialization version id

🟥 Not needed

[codecov]

Codecov Report

Attention: Patch coverage is 91.66667% with 13 lines in your changes missing coverage. Please review.

Project coverage is 69.70%. Comparing base (03dae4e) to head (fd33464).

Files	Patch %	Lines
...m/filterchain/ItineraryListFilterChainBuilder.java	35.71%	8 Missing and 1 partial ⚠️
...rithm/mapping/RouteRequestToFilterChainMapper.java	0.00%	1 Missing and 1 partial ⚠️
...iority/TransitGroupPriorityItineraryDecorator.java	0.00%	1 Missing ⚠️
...orithm/filterchain/filters/system/mcmax/State.java	98.78%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@              Coverage Diff              @@
##             dev-2.x    #5919      +/-   ##
=============================================
+ Coverage      69.65%   69.70%   +0.05%     
- Complexity     17156    17224      +68     
=============================================
  Files           1942     1947       +5     
  Lines          73786    73939     +153     
  Branches        7550     7573      +23     
=============================================
+ Hits           51397    51541     +144     
- Misses         19759    19765       +6     
- Partials        2630     2633       +3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[t2gran]

This PR is improving the group-by filter, reminding improvements is documented her #5923

[leonardehrenfried]

I typical Thomas fashion this is using very clean coding practises and is well tested and documented. Nevertheless, I picked up on a few smaller issues.

[t2gran]

The merge does not look ok - I will try to do it one more time...

[leonardehrenfried]

You apparently formatted all files in the client folder. I'm sure that wasn't intentional.

[t2gran]

I noticed, probably did it as part of comitting from IntelliJ. Did the merge again, but now using git from the command line.

[leonardehrenfried]

What does it mean to "merge in a group id" rather than merging two group ids? I would assuming that "merging in" modifies an existing instance.

[t2gran]

The first argument is a set of group ids, not singular.

[t2gran]

I guess both names work, but the name should match the JavaDoc:

/**
   * Merge in the transit group id with an existing set. Note! Both the set
   * and the group id type is {@code int}.
   *
   * @param currentGroupIds the set of groupIds for all legs in a path.
   * @param boardingGroupId the transit group id to add to the given set.
   * @return the new computed set of groupIds
   */

[leonardehrenfried]

I have a question about the renaming of "mergeInGroupId" method.