perf: optimize refcounting

[yannham]

Depends on #2455.

This commit mostly copies the style of the stdlib in std::rc::Rc, in hope of gaining the same kind of optimizations. It does nullify the loss of #2455, and wins overall 5-7% on the OPL benchmark in runtime when compared to master. After a bit of experimentation, it seems that the [u8; 7] representation had to go for those optimizations to fire properly.

[github-actions]

Bencher Report

Branch	perf/optimize-refcounting
Testbed	ubuntu-latest

Click to view all benchmark results

Benchmark	Latency	microseconds (µs)
diagnostics-benches/inputs/goto-perf.ncl	📈 view plot 🚷 view threshold	10,096.00 µs
diagnostics-benches/inputs/large-record-tree.ncl	📈 view plot 🚷 view threshold	185,390.00 µs
diagnostics-benches/inputs/redis-replication-controller.ncl	📈 view plot 🚷 view threshold	306.37 µs
diagnostics-benches/inputs/small-record-tree.ncl	📈 view plot 🚷 view threshold	429.21 µs
fibonacci 10	📈 view plot 🚷 view threshold	218.69 µs
foldl arrays 50	📈 view plot 🚷 view threshold	369.50 µs
foldl arrays 500	📈 view plot 🚷 view threshold	2,669.30 µs
foldr strings 50	📈 view plot 🚷 view threshold	2,827.50 µs
foldr strings 500	📈 view plot 🚷 view threshold	26,977.00 µs
generate normal 250	📈 view plot 🚷 view threshold	21,603.00 µs
generate normal 50	📈 view plot 🚷 view threshold	1,155.60 µs
generate normal unchecked 1000	📈 view plot 🚷 view threshold	1,650.40 µs
generate normal unchecked 200	📈 view plot 🚷 view threshold	335.92 µs
init-diagnostics-benches/inputs/goto-perf.ncl	📈 view plot 🚷 view threshold	52,824.00 µs
init-diagnostics-benches/inputs/large-record-tree.ncl	📈 view plot 🚷 view threshold	205,430.00 µs
init-diagnostics-benches/inputs/redis-replication-controller.ncl	📈 view plot 🚷 view threshold	47,784.00 µs
init-diagnostics-benches/inputs/small-record-tree.ncl	📈 view plot 🚷 view threshold	47,699.00 µs
pidigits 100	📈 view plot 🚷 view threshold	1,890.30 µs
pipe normal 20	📈 view plot 🚷 view threshold	575.16 µs
pipe normal 200	📈 view plot 🚷 view threshold	4,299.70 µs
product 30	📈 view plot 🚷 view threshold	259.95 µs
requests-benches/inputs/goto-perf.ncl-000	📈 view plot 🚷 view threshold	3,018.50 µs
requests-benches/inputs/large-record-tree.ncl-000	📈 view plot 🚷 view threshold	585,160.00 µs
requests-benches/inputs/large-record-tree.ncl-001	📈 view plot 🚷 view threshold	88.70 µs
scalar 10	📈 view plot 🚷 view threshold	489.00 µs
sum 30	📈 view plot 🚷 view threshold	261.23 µs

🐰 View full continuous benchmarking report in Bencher

[dpulls]

🎉 All dependencies have been resolved !