the glue dries differently, depending on the proportions of the ingredients, at the beginning I suggest a little less hardener
I would not advise grinding the grooves because the accuracy must be a few micrometers, preferably 1

They won't have them in teps, a team with a suitcase arrives there and welds, or they use disconnectors. And even if they make them separate, they prefer factory ones with a piece of optical fiber and weld them together.
Try a ferrule

It's less work than with the groove. Instead of adjusting so that they lie evenly in the grooves, you simply slide it into the tube and seal it

The trick in both cases is to cut the optical fiber well, to a great extent a technique similar to glass cutting. If you have a new knife, no problem. The face must be perpendicular to the axis.
There are no expensive gauges. The reflectometer immediately falls off. As already suggested by the level meter
This one is probably expensive but more or less described for what
http://www.tomtronix.com.pl/avo/mpm1000.htm The price of the source is $1000, the price of the meter with the built-in source is $1500 (one window) $2000 (two windows), so maybe he can convince the boss ;)
It would be best to connect and see if it works. If it doesn't work, the only way out is to get a meter and measure both sections of the track, because you don't know if you glued it wrong or there are surprises elsewhere.
V grooves are expensive to produce (good quality) and hard to get a good result because a lot depends on the fastener, dirt and dexterity. It's easier with SC.
PS However, I would ask for optijacks, they were designed for such needs. Only they were introduced 5 years ago and they are having trouble breaking through the standards