Removes torch dataloader from load_data, adds initial tests to dataloading

[ssenan]

• Resolve torch dataloader tied to preprocessing function #155
• Moves the torch dataloader instantiation to train_util class
• Adds tests to SequenceDataset class that is used to instantiate dataloader
• Adds hatch run test to build workflow

[codecov]

Codecov Report

Merging #156 (c329776) into main (a55002d) will increase coverage by 8.66%.
The diff coverage is 69.23%.

@@            Coverage Diff            @@
##            main     #156      +/-   ##
=========================================
+ Coverage   2.73%   11.39%   +8.66%     
=========================================
  Files         12       13       +1     
  Lines        805      825      +20     
  Branches      89       89              
=========================================
+ Hits          22       94      +72     
+ Misses       783      730      -53     
- Partials       0        1       +1     

Impacted Files	Coverage Δ
src/dnadiffusion/data/dataloader.py	26.92% <0.00%> (+26.92%)	⬆️
src/dnadiffusion/utils/train_util.py	0.00% <0.00%> (ø)
tests/test_dataloader.py	100.00% <100.00%> (ø)

... and 1 file with indirect coverage changes

[cameronraysmith]

Perhaps the tests would be run twice with the suggested change to the build workflow.

[cameronraysmith]

Since the two adjacent lines here are the same,

DNA-Diffusion/pyproject.toml

Line 141 in a55002d

test-cov-xml = "pytest -rA --cov-report=xml"

this should be redundant (i.e. the new functionality would essentially run the tests twice).