Fix constant compression in-place check for bools

[benjaminwinger]

Bools are stored packed in column chunks, so indexing into the data using the size in bytes of the type will not work.

This is done as a quick fix, however I think it would be a good idea to move away from passing around raw uint8_t* when passing around type-generic data. Instead, I think we should create abstractions for storing and accessing the raw buffers. That would let us more easily distinguish between, for example, a ValueVector buffer and a ColumnChunk buffer (e.g. the former stores 1 bool per byte, and the latter 1 bool per bit), and also make sure we handle these edge cases.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.27%. Comparing base (fa0ef79) to head (e35aec3).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3211      +/-   ##
==========================================
+ Coverage   92.26%   92.27%   +0.01%     
==========================================
  Files        1161     1161              
  Lines       44129    44143      +14     
==========================================
+ Hits        40714    40732      +18     
+ Misses       3415     3411       -4     

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

[benjaminwinger]

Apparently this issue was already handled separately for Nulls (which this change initially broke). I don't really like that the implementations diverged.
I've removed some of the NullColumn functions which have become almost identical to the base Column versions.

[bigluck]

Thanks @benjaminwinger