Optimize gtf parsing

[cmdcolin]

There is a nice ability to load GTF files directly because hosts like UCSC still publish these for their annotations

e.g.
https://hgdownload.soe.ucsc.edu/goldenPath/canFam6/bigZips/genes/ncbiRefSeq.gtf.gz

this 19MB file does load sometimes but pushes memory limits

Here is a partial profiling trace of loading the url above (not a full trace because leaving it on during loading crashes on my computer)

zoom out of loading process
gtf_loading.json.gz

zoom in

[cmdcolin]

this is the memory line going up to 800+mb

[cmdcolin]

cc @teresam856 could be worth checking out. the 19mb gz file does unpack into 329mb in memory but it would cool if we could handle this scale of file.

[cmdcolin]

the 329 is just checking size on disk from unzipping the gz

[teresam856]

This is a good thing to have. Could look into it...

[cmdcolin]

super..if it was optimized, this could become a "recommended" or even pre-configured way for people to load data

[cmdcolin]

some amount better from #2927