Closes #31 (closed) and #32 (closed). Since !72 (merged), where Retina clusters became the default for Allen, the tests have all run over files with Retina clusters. However, there are still plenty of examples with no Retina clusters, which were untested and broken by this change. This fixes them by using the
hlt1_pp_veloSP sequence, and adds efficiency and rate tests on the
- Removed the redundant/duplicated
hlt1_rate_default.pyis exactly the same,
- Remove the
no_gecexamples with no retina clusters, since there is no such Allen sequence. These files were broken and not tested since !72 (merged),
- Remove an old TOS efficiency test: all the efficiency tests now test this functionality,
- Reveal some hidden seg faults in the tests from an erroneously caught exception in
- Then patch those seg faults: the wizard options template needs all the retina-cluster based boilerplate if using the default Allen sequence,
- Make a bit of effort to use more descriptive tuple paths, partly for clarity and partly to avoid different tests writing/reading to the same place when multithreaded,
- Try a little to improve the alignment and truncation of rates results (closes #32 (closed)).
Catch #32 (closed)?
Moore MR on documentation
Kill the no-gec-no-retina-cluster examples?
Test for non-retina cluster examples
Hopefully avoid hard-coding the bank types forPostpone.