[SYSTEMDS-3548] Optimize IO path Python interface for SystemDS

[Nakroma]

Student project SYSTEMDS-3548 and follow-up to #2154

Contributions/discussion:

• I did some follow-up to both suggestions from @Baunsgaard in the first PR and did testing with both chunking into smaller parts and fusing operations into fewer java calls. I was unable to get any real improvements that were replicable over the larger datasets, altough I don't have a ton of experience with Py4J, so this might still have some potential. I added some of the adjacent code for it though (fusing convert, setting only chunks of a FrameBlock etc.), so at least some of the work I did there contributes to the project.
• As it turns out, anything involving the java gateway is super costly, so for example even simply doing a if var == jvm.gate.sds.ValueType.String comparison has a big overhead. I was able to optimize another constant time by reducing stuff like that to a minimum, see the first graph below.
• For cases where cols > rows the current column-wise processing is very slow, so I added row-wise processing for that case to speed it up (see second graph below, tested on 1k rows x 10k cols). Note here that it currently only does that for edge cases where all columns have the same data type. This is because when testing, serializing over a row with different columns was very costly. I wasn't able to spend a lot of time on this, as the deadline is approaching, so I think there is a lot of potential here to find an efficient way to serialize to be able to also use it for mixed columns. I'd also expect in the most optimal case for the time to be the same as as the rows > cols case, so I think there is probably also more optimization potential in the current row-wise processing.
• Small note that I switched how I compared times, before I was averaging runs but now I take the min as suggested by the timeit docs, so times might be slightly different from the first PR.
• Fixed a regression from my first PR where exceptions in the threaded function calls wouldn't propagate properly.
• Fixed a small bug in the perftests to be able to read multi-file data (since that's what datagen generates for larger datasets)

[Baunsgaard]

LGTM, thanks for the edits,

The only elements missing are documentation on the methods defined. If you could add these, then i will merge it!

[codecov]

Codecov Report

Attention: Patch coverage is 84.31373% with 8 lines in your changes missing coverage. Please review.

Project coverage is 72.27%. Comparing base (f8522a7) to head (cac793e).
Report is 8 commits behind head on main.

Files with missing lines	Patch %	Lines
...rg/apache/sysds/runtime/frame/data/FrameBlock.java	63.63%	5 Missing and 3 partials ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #2189      +/-   ##
============================================
+ Coverage     71.88%   72.27%   +0.38%     
- Complexity    44701    44981     +280     
============================================
  Files          1449     1452       +3     
  Lines        169182   169330     +148     
  Branches      32980    33032      +52     
============================================
+ Hits         121617   122383     +766     
+ Misses        38237    37637     -600     
+ Partials       9328     9310      -18     

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

[Baunsgaard]

Github actions are bringing up a valid point, testing is missing.

We need to add a few test files (with tests).

1. in Python

• Add files to src/main/python/tests/iotests/???.py for end-to-end tests.

3. in Java, Please extend:

• src/test/java/org/apache/sysds/test/component/frame/array/Py4jConverterUtilsTest.java
• And add a src/test/java/org/apache/sysds/test/component/matrix/???.java if you find it appropriate.

[Nakroma]

Added those. I've added the FrameBlock tests to /frame/FrameGetSetTest.java and the python tests to the already existing pandas setup since that made the most sense to me, let me know if thats okey if not I can split that up into their own files.