Slotted enumeration faster

[Scooletz]

This PR introduces two changes to enumeration over SlottedArray:

1. EnumerateAll that enumerates now with a bit less overhead (code size reduction and some inlines)
2. EnumerateNibble that uses the new encoding of length to to leverage the filtering and returns data without having their keys unprepared if not needed. It's not 16 faster, but gives ~3x performance boost for scans.

Benchmarks

For EnumerateAll the existing one is used. For EnumerateNibble new benchmarks were added as previously it was filtering over the materialized key which was not different from EnumerateAll.

Before

Method	Mean	Error	StdDev	Code Size
EnumerateAll	751.6 ns	7.62 ns	6.75 ns	1,913 B

After

Method	nibble	Mean	Error	StdDev	Code Size
EnumerateAll	?	744.7 ns	8.57 ns	8.01 ns	1,883 B
EnumerateNibble	0	239.3 ns	2.72 ns	2.54 ns	1,939 B
EnumerateNibble	1	241.3 ns	4.42 ns	4.14 ns	1,932 B

[github-actions]

Package	Line Rate	Branch Rate	Health
Paprika	85%	79%	➖
Summary	85% (4307 / 5088)	79% (1353 / 1723)	➖

Minimum allowed line rate is 75%