fix for lottery choice bug for empty alts table

[mxndrwgrdnr]

This PR fixes a bug initerative_lottery_choices() which creates an empty merged choice table object if/when all rows of alts table are filled and dropped before max capacity is exceeded.

[github-actions]

Test coverage is 76%

[waddell]

Thank you, @mxndrwgrdnr. UDST has resumed active maintenance of ChoiceModels, and this focused bug fix is now tracked under #78.

The failure mode is plausible: after the alternatives table has been exhausted, the loop should terminate before attempting to construct another merged choice table. Before integration, the change needs a regression test that exhausts the alternatives and verifies both clean termination and the expected unallocated-chooser behavior. We will also run it against the modern compatibility baseline once that CI work is in place.

If you are available and would like to add the test after the baseline is ready, that would be welcome. Otherwise, UDST can adapt the fix with the original authorship and explicit credit preserved.