Consolidate hypothesis testing 178390751

[ernestoarbitrio]

Notion page: https://www.notion.so/crunch/Sort-by-value-Transform-for-Dashboards-and-Export-PRD-759062891aa2471e8f9ca9c766ee6e72#440ab7f0a4234e8eb811678762ffb17a

Goal: support sort-by-value for pairwise significance measures.

All pairwise significance measures have been moved from methods to lazyproperties.

The new way to specify the pairwise sig column baseline is to use a new object in the transforms:

        transforms = {
            "columns_dimension": {
                "pairwise_significance": {
                    "column_significance": {"element_id": 0}  # < - normal column
                }
            }
        }

        transforms = {
            "columns_dimension": {
                "pairwise_significance": {
                    "column_significance": {
                        "insertion_id": 1 or "str_id"  # < - insertion column
                    }
                }
            }
        }

[codecov]

Codecov Report

Merging #320 (ea06064) into master (80986d5) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##            master      #320    +/-   ##
==========================================
  Coverage   100.00%   100.00%            
==========================================
  Files           20        19     -1     
  Lines         4639      4848   +209     
==========================================
+ Hits          4639      4848   +209     

Impacted Files	Coverage Δ
src/cr/cube/cube.py	100.00% <100.00%> (ø)
src/cr/cube/cubepart.py	100.00% <100.00%> (ø)
src/cr/cube/dimension.py	100.00% <100.00%> (ø)
src/cr/cube/matrix/assembler.py	100.00% <100.00%> (ø)
src/cr/cube/matrix/measure.py	100.00% <100.00%> (ø)

---

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 80986d5...ea06064. Read the comment docs.

[scanny]

Okay, I'm out of time for today, but here are some to get you started.

[scanny]

Okay, I didn't get far when I discovered a basic error in how this is organized. Indices need to be computed in the measure and not in assembler. Then assembler's role is just to adjust the indices from payload-order to sorted-order. Everything else (like ._assemble_matrix happens the same way as any other measure.

[scanny]

Okay, LGTM after some docstring improvement :)

[gergness]

I had originally wanted to go through and understand the whole thing, which I haven't had time to do yet. (Not sure what you were asking for, but since Steve already approved, is it good?)

One comment about a docstring change

[gergness]

docstring is out of date now, right?
"Optional int index" -> "Optional element id" (which is either an int/str if you want to have the type)

(I think Mike's arguing in slack that this should be called the "code", but this would be the only place we use that terminology)