Refactor fragment_iterator()

[ns-rse]

Currently we have...

    def fragment_iterator(read_iterator):
        read_list = list()
        last_read = None

        for read in read_iterator:
            if last_read is not None and last_read != read.query_name:
                yield read_list
                read_list = list()
                last_read = read.query_name
            last_read = read.query_name
            read_list.append(read)

        yield read_list

Where read_iterator is the bamfile.

Need to think through the logic here......

1. last_read always starts as None so on the first pass last_read is set to the current read.query_name and
   nothing is run. What does this first read look like, do we need it?
2. Loop then runs unless read.query_name matches last_read, yielding read (as a list) which will actually be the
   value from the previous pass.

Following switches to passing the bam_file path, adds typehints and docstring.

from isoslam import io

def fragment_iterator(bam_file: str | Path) -> Generator[?, None, None]:
    """
    Iterate over fragments from a `.bam` file.

    Parameters
    ----------
    bam_file : str | Path

    Returns
    -------
    Generator
        Iterable of reads from BAM file.
    """
    ...
    for read in io.load_file(bam_file):
        ...

I need to work through what this is doing and get a better understanding. Tests here we come!