refactor: use dataclass for structured tsv reads

[lavafroth]

Changes

• Create a TSV structure to hold start, size and strand.
• Bind the current read to a variable instead of repeatedly indexing into support.
• Create a slice to extract left and right sequences.

[readmanchiu]

I've been leery of the among of memory used to create an object for every support read instead of a simple tuple.

[lavafroth]

When the code gets compiled to bytecode, the class (struct) has three fields, so the memory consumption should very likely reduce. Also, the slice objects consume the same memory as a tuple and the changed code will improve readability without incurring a performance penalty.

[readmanchiu]

Traceback (most recent call last):
  File "/projects/btl_scratch/rchiu/tmp/straglr/extract_repeats.py", line 7, in <module>
    from dataclasses import dataclass
  File "/home/rchiu/miniconda2/lib/python3.9/dataclasses.py", line 5, in <module>
    import inspect
  File "/home/rchiu/miniconda2/lib/python3.9/inspect.py", line 39, in <module>
    import importlib.machinery
  File "<frozen importlib._bootstrap>", line 1007, in _find_and_load
  File "<frozen importlib._bootstrap>", line 986, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 680, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 786, in exec_module
  File "<frozen importlib._bootstrap_external>", line 881, in get_code
  File "<frozen importlib._bootstrap_external>", line 980, in get_data

[lavafroth]

You might have conflicting modules, maybe a file with the same name. Dataclasses were introduced in Python 3.7