fix for writing large_fasta sequences

[epaule]

• the window count was off for cases where the sequence length aligned with the window size
• i also bumped up the max buffer size

[hyphaltip]

I haven't tested this yet myself but this seems like an appropriate update - is there a performance improvement or a particular bug this fixes?

[epaule]

it fixes a bug: namely that it tried to write past the end of the sequence in cases where the number of base pairs of the last window aligned with the window. It then tried to write bp end+1 to end.

[cjfields]

Awesome, thanks for the addition @epaule !