The percentage of the market that holds a strong opinion about a nondefault web browser with already minuscule (and consistently declining) market share, and will actively avoid you for not supporting it, almost definitely costs more to cater to then to ignore. (Testing time, engine specific bugs and code, etc.)

That small percentage will be extremely loyal to whichever services will cater to them though. A small percentage of users all choosing to use your service can still be quite a lot of users if there is no real competition in the segment.