FirstValue / LastValue performance optimisations

[blaginin]

Is your feature request related to a problem or challenge?

FirstValue and LastValue can be faster!

1. When ignore nulls is passed, we sort groups, although getting just the first/last non-null value would be enough.
2. Without ignore nulls, there's still some small overhead that can be removed. For example, creating a row buffer for each row in our group.
3. For some primitive types, we can implement GroupsAccumulator similarly to Min/Max.

Describe the solution you'd like

No response

Describe alternatives you've considered

No response

Additional context

No response

[blaginin]

take