Add different dict modes to compression ratio regression test, update results.csv

[senhuang42]

On the row-hash PR, DMS was silently broken until recently fixed/discovered - our coverage of tests on different dict modes (DMS, DDS, Copy, Load) is not particularly high. This should be a good first step in ensuring we detect such things. (The row-hash PR adds a DMS unit test in fuzzer.c as well, to check for regressions)

This PR:

• Adds ratio regression tests to clevels >= 1 for each dictMode when testing using advanced API that can honor requested parameters.
• Updates results.csv
  • New rows with new dictmode tests
  • Adjustments for improvements introduced with block splitter and levels 16, 19
  • Minor spelling fix
• Fixes compiler warnings when building this regression test

Testing: I tested this on row-hash with a broken DMS, and indeed the compression results were bad.

[senhuang42]

So I added the new configs to LEVEL() so we automatically get testing for all the levels we try.

An alternative approach is to have each of these just be their own config, with a fixed compression level (and if we wanted to test more clevels/strategies, we could just add more configs). But that does have the downside of not automatically hitting all the tested clevels.

I do prefer having these run on all the compression levels, in case in the future, strategies get changed in a way that might affect the dictionary strategies.

[Cyan4973]

makes sense

[terrelln]

Thanks for adding this Sen! Really glad to see increased coverage of dictionary compression.