Accounting for end: false in the matcher score

[skirtles-code]

While working on #2148, I ran into some problems with end: false.

I was trying to ensure that my changes didn't impact the matching of routes, even in weird edge cases. In most cases that proved relatively straightforward, apart from for end: false.

My underlying assumption is that the route ranking mechanism is supposed to ensure that 'more specific' routes come before 'less specific' routes. Generally speaking, that does seem to be how it's implemented.

My PR handles static strings in paths specially. As static strings are 'more specific' than dynamic strings, that usually works out fine, as they'd come earlier in the ranking anyway.

But end: false is a notable outlier. It behaves a lot like a .* wildcard, but it doesn't really get taken into account when calculating the score. It does impact the inclusion of the 'strict bonus' for strict: true, but apart from that it's ignored.

With enough work, it probably is possible to get #2148 to retain the current sort order for matchers, but it'd likely involve a lot of special handling just for end: false. Instead, I think it would be better to change the existing behaviour, so that end: false is taken into account in the score.

If this idea is acceptable in principle, I'm happy to have a go at implementing it in pathParserRanker.ts. This would be separate from #2148, though it should ultimately allow that PR to be completed.

[posva]

I actually wonder if anybody uses end: false. I initially kept it because Vue Router 3 had it in path-to-regexp but I don't find it useful by default. I think that it could be its own matcher (once it's allowed to pass a matcher to the router creation)