Pool transitions buffers to reduce allocations

[sayrer]

This change adds a dfaTransitions field to nfaBuffers and uses it in valueMatcher.transitionOn() instead of allocating a new slice for each value match operation. The buffer is reset to length 0 before each use while preserving capacity. See #470.

Reliable reduction in allocations.

Before:

 % go test -bench=^Benchmark8259Example$ -run=^$ 
FA: Field matchers: 2 (avg size 2.500, max 4)
Value matchers: 5
SmallTables 20371 (splices 6, avg 4.033, max 66, epsilons avg 0.001, max 2) singletons 1
97182/sec
goos: darwin
goarch: arm64
pkg: quamina.net/go/quamina
cpu: Apple M1 Ultra
Benchmark8259Example-20    	  127725	     10290 ns/op	     984 B/op	      31 allocs/op
PASS
ok  	quamina.net/go/quamina	1.620s

After:

% go test -bench=^Benchmark8259Example$ -run=^$                    
FA: Field matchers: 2 (avg size 2.500, max 4)
Value matchers: 5
SmallTables 20371 (splices 6, avg 4.033, max 66, epsilons avg 0.001, max 2) singletons 1
100830/sec
goos: darwin
goarch: arm64
pkg: quamina.net/go/quamina
cpu: Apple M1 Ultra
Benchmark8259Example-20    	  130026	      9918 ns/op	     914 B/op	      23 allocs/op
PASS
ok  	quamina.net/go/quamina	1.628s

[timbray]

I like PRs that only touch 2 files. Tbh I'm surprised at the size of the effect. Good stuff!