Optimize reserve for non-unique EcoVec

[ndren]

This PR avoids bounds checking when calling reserve on a non-unique EcoVec when it has to clone over all the values to a new EcoVec.

Some results, before/after:

size=   100, reserve=   500:       1593 ns/iter
size=  1000, reserve=  5000:       9057 ns/iter
size= 10000, reserve= 50000:      89608 ns/iter
size=100000, reserve=500000:     895274 ns/iter

size=   100, reserve=   500:        637 ns/iter
size=  1000, reserve=  5000:       5752 ns/iter
size= 10000, reserve= 50000:      51176 ns/iter
size=100000, reserve=500000:     513765 ns/iter

[laurmaedje]

Thanks!