Add second criteria to the Raptor heuristics

[t2gran]

Summary

This PR fixes a bug in Raptor's destination pruning heuristics where the secondary cost criterion (c2) was not being considered when evaluating whether paths can be pruned. When using multi-criteria search with transit priority groups, this could cause paths to be incorrectly pruned based on incomplete cost information (only c1, ignoring c2). Note! There
are usually a lot of slack in the heuristics due to the estimation of c1, so this bug only apply when the slack is less than the c1 relaxation.

Issue

Closes #7112

Raptor Transit Routing

The HeuristicsProvider creates optimistic best-case estimates for paths to determine if continuing a search branch is worthwhile. It does this by combining actual arrival state with pre-calculated heuristics to estimate the best possible final destination arrival. These estimates are then checked against the current best known paths using DestinationArrivalPaths.qualify().

Before this fix:

• HeuristicsProvider.qualify() was only passing c1 (primary cost) to DestinationArrivalPaths.qualify()
• The c2 value (secondary cost, used for transit priority) was hardcoded to NOT_SET in dummy paths
• This meant pareto comparison during pruning didn't account for transit priority penalties

After this fix:

• c2 is now threaded through the entire qualify chain
• HeuristicsProvider.qualify() → DestinationArrivalPaths.qualify() → Path.dummyPath() all pass c2
• Destination pruning now correctly considers both cost criteria

Testing

• Updated PathTest to include c2 values in test cases, verifying path comparison respects c2 ordering
• Added a module test I02_C2DestinationPruningTest that specifically tests destination pruning with transit priority groups
• Test verifies that paths are correctly qualified/disqualified based on c2 values when using multi-criteria search

Performance Impact

This change modifies Raptor's heuristics evaluation, but only adds parameter passing - no additional computation. The qualify() method was already being called; it now has access to complete cost information. This does however increase the number of paths evaluated, so a smal increase in the search time is expected - now including possible optimal paths, witch was dropped earlier.

Performance verification

I have not seen any significant changes in the SpeedTest run on the Norway data set:
Before

 ==> multi_criteria_destination : [  294,  280,  277,  279,  280 ] Avg: 282,0  (σ=6,1)
 ==> multi_criteria_destination : [  420,  389,  386,  390,  390 ] Avg: 395,0  (σ=12,6)

After fix

 ==> multi_criteria_destination : [  296,  282,  291,  281,  282 ] Avg: 286,4  (σ=6,0)
 ==> multi_criteria_destination : [  426,  393,  403,  392,  391 ] Avg: 401,0  (σ=13,2)

Documentation

✅ Java Documentation updated - the modified methods maintain clear javadoc

Changelog

✅ Should be included in changelog

Bumping the serialization version id

🟥 No serialization changes

This PR modifies Raptor's internal routing logic but doesn't change any serialized graph data structures or edge properties. No serialization version bump needed.

[leonardehrenfried]

You have a conflict.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 72.17%. Comparing base (2540a97) to head (f70f649).
⚠️ Report is 3 commits behind head on dev-2.x.

Additional details and impacted files

@@            Coverage Diff             @@
##             dev-2.x    #7165   +/-   ##
==========================================
  Coverage      72.17%   72.17%           
- Complexity     20877    20878    +1     
==========================================
  Files           2273     2273           
  Lines          84437    84438    +1     
  Branches        8424     8424           
==========================================
+ Hits           60942    60945    +3     
+ Misses         20520    20518    -2     
  Partials        2975     2975           

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

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.