Support nested structs in rank and dense rank

[rwlee]

Follow on to #8652 for nested struct support using, partially removing the need for #8683.

This change simplifies the rank algorithm by assuming superimpose_parent_nulls has been ran on the struct column. This removes the need for separate logic that ensures we are not comparing elements covered by a parent column's null mask.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (branch-21.10@c6ddd46). Click here to learn what that means.
The diff coverage is n/a.

@@               Coverage Diff               @@
##             branch-21.10    #8962   +/-   ##
===============================================
  Coverage                ?   10.82%           
===============================================
  Files                   ?      115           
  Lines                   ?    19166           
  Branches                ?        0           
===============================================
  Hits                    ?     2074           
  Misses                  ?    17092           
  Partials                ?        0           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c6ddd46...eb5af8a. Read the comment docs.

[rwlee]

Moving back to draft because of parent null inheritance assumption concerns. Will rework around changes to the flattening approach and null checking process.

[hyperbolic2346]

I like how this cleaned up and boiled down to basically the same pattern for all of these.

[vyasr]

Nice, it is much cleaner now that you've pushed the null handling down to the comparator. Would it be possible to change generate_ranks and generate_dense_ranks into a single internal function that accepts a device lambda for the tabulate and a predicate for the inclusive_scan_by_key? I believe that the rest of the code is entirely identical now.

[vyasr]

Changes look great. Nice to see that the templating made this a negative LOC PR.

[harrism]

cmake approval

[harrism]

@gpucibot merge