Leverage reciprocal_exists for handling reciprocal fusions in OncoPrints

[jaclyn-taroni]

Related to #1009 - We were having an issue where reciprocal fusions were showing up inappropriately as multi-hit fusions in the OncoPrints. During review, using reciprocal_exists in the putative oncogenic fusion file was raised https://github.com/AlexsLemonade/OpenPBTA-analysis/pull/1009#pullrequestreview-647703181 which I came across during my own review of #1009. It occurred to me that we only need to sort gene partners of fusions in alphabetical orders when we know a reciprocal fusion exists, where we then use the sorted results to collapse reciprocal fusions/make sure we're only counting them once.

So here I'm proposing fusion code updates that do the following:

• Handle reciprocal_exists == TRUE and reciprocal_exists == FALSE separately, based on the considerations for each (e.g., sorting for reciprocal_exists == TRUE only) - I draw from the sorting code in #1009 but use stringr functionality because personal preference, mostly
• Binds the reciprocal and no reciprocal fusion data frames
• Handles multi-hit fusions the same way as we do in master

@cbethell I spot checked a couple differences between this code and what's in #1009 and it looks like this code correctly IDs some multi-hit fusions missed in #1009.

[jaclyn-taroni]

Because I'm not merging into master, no review required. Feel free to merge if this looks good, @cbethell !

[cbethell]

Because I'm not merging into master, no review required. Feel free to merge if this looks good, @cbethell !

Took a careful look at things locally and everything looks good ✅ so I'll go ahead and merge (then address and re-run things over on #1009)