Completed a TODO on barrier free SkipList::Insert() and cleaned some unused headers

[FlavioMili]

I tested the change using perf on previous main version and my branch

perf record -F 999 -g -- ./db_bench_old
mv perf.data perf_old.data

perf record -F 999 -g -- ./db_bench_new
mv perf.data perf_new.data

From the results I got a consistent improvement of up to 10% on the Insert() function.

I needed to add a second function because the other option I had was to use templates with two functions anyway and to me it looked harder to read for a newcomer or maintainer.
There was also the option to add a ternary operator inside the loop but it ended up having no improvement at all, or even being slightly worse.

Environment

• CPU: 16 * AMD Ryzen AI 7 350 w/ Radeon 860M
• Cache: 1024 KB
• Keys/Values: 16B keys / 100B values (1M entries)

Performance Comparison (micros/op)

I would still take these with a grain of salt, some results are too close to be compared in my opinion

| Benchmark | Baseline (Old) | Optimized (New) |
| :--- | :---: | :---: | :---: |
| fillrandom | 1.789 | 1.745 |
| overwrite | 2.474 | 2.419 |
| fill100K | 418753 | 416100 |
| fillsync | 3.591 | 3.570 |
| readrandom | 5.338 | 5.235 |
| readseq | 0.182 | 0.183 |