1) there is only one true "battery locomotive" on US rails right now, and it's back at the WABTEC (GE) factory right now as data from its test season with BNSF are analyzed. IIRC it carried about 45 MWH of batteries. Not sure what chemistry was used. For comparison, a long-range Tesla battery holds about 100 KWH. Note the order of magnitude.

2) it strongly depends on the type of Li battery used. Some chemistries are more subject to fires than others. The Li-Co-usually Mn chemistries used in cars are more susceptible because, for reasons of packaging and the need for high energy density, high power, small size, and (relatively) light weight are priorities. The tendency for the batteries to overheat and go into thermal runaway is aggressively, and usually successfully, designed around with cooling systems, special internal designs, and battery management systems. Li-Iron-Phosphate batteries are much less susceptible to thermal runaway and fire than typical automotive batteries, and in fact are almost considered fireproof; though they store less energy per unit mass, and are becoming the standard for "solar batteries" and similar setups where you Really Don't Want the battery catching fire at some random time.

3) IIRC, Tesla's stationary batteries until recently were essentially a whole bunch of car batteries installed in boxes. As with car batteries, if something goes wrong and one of the modules catches fire, it quickly overheats its neighbors, and pretty soon you have a general conflagration. It's very hard to put out - in fact, you pretty much can't. Fire attack on a Li battery mainly floods it with lots of water to try to cool it to where the fire goes out on its own. Not the kind of battery I want in my garage deep-cycling every day to keep my solar panels from having to deal with net metering (or the lack of it), or as a local storage module in a utility substation. For a home battery, I'd rather have Li-Fe (commonly abbreviated as LiFP or LFP) - much safer, and in a stationary application you can make up for the lower energy density by using more of them (taking up more space).

Back to battery locomotives. WABTEC does want to try it again. The battery unit, combined with diesels, produces a useful hybrid locomotive *consist* (rather than a single hybrid unit, which is complicated, heavy, and expensive). One article I saw said that they figure it'll need at least 60 MWH of storage to do the job well. My guess is that they'll also want to use LFP if they can, for the safety aspects as well as the longer life compared to automotive Li-Co-Al or Mn chemistries.