fix(virtq/split): don't box TransferTokens

[mkroening]

This was proposed by @hcsch in #1995 (2efa9ca).

Does that make sense without that PR's context? Can we measure any difference here?

[github-actions]

Benchmark Results

Benchmark	Current: d794594	Previous: 6019ba4	Performance Ratio
startup_benchmark Build Time	110.70 s	110.61 s	1.00
startup_benchmark File Size	0.89 MB	0.91 MB	0.99
Startup Time - 1 core	0.90 s (±0.03 s)	0.93 s (±0.03 s)	0.97
Startup Time - 2 cores	0.90 s (±0.02 s)	0.93 s (±0.04 s)	0.97
Startup Time - 4 cores	0.90 s (±0.03 s)	0.94 s (±0.03 s)	0.96
multithreaded_benchmark Build Time	111.08 s	110.93 s	1.00
multithreaded_benchmark File Size	1.00 MB	1.01 MB	0.99
Multithreaded Pi Efficiency - 2 Threads	88.01 % (±9.61 %)	92.34 % (±7.09 %)	0.95
Multithreaded Pi Efficiency - 4 Threads	43.05 % (±3.84 %)	44.95 % (±3.07 %)	0.96
Multithreaded Pi Efficiency - 8 Threads	25.00 % (±2.41 %)	25.64 % (±1.60 %)	0.98
micro_benchmarks Build Time	269.97 s	273.99 s	0.99
micro_benchmarks File Size	1.01 MB	1.01 MB	0.99
Scheduling time - 1 thread	161.38 ticks (±13.61 ticks)	162.90 ticks (±21.02 ticks)	0.99
Scheduling time - 2 threads	95.03 ticks (±13.20 ticks)	89.13 ticks (±11.93 ticks)	1.07
Micro - Time for syscall (getpid)	8.56 ticks (±4.63 ticks)	10.81 ticks (±4.73 ticks)	0.79
Memcpy speed - (built_in) block size 4096	58385.38 MByte/s (±42276.13 MByte/s)	64709.21 MByte/s (±45699.60 MByte/s)	0.90
Memcpy speed - (built_in) block size 1048576	15783.38 MByte/s (±12997.93 MByte/s)	15408.89 MByte/s (±12426.51 MByte/s)	1.02
Memcpy speed - (built_in) block size 16777216	11400.25 MByte/s (±9274.77 MByte/s)	11558.07 MByte/s (±9407.44 MByte/s)	0.99
Memset speed - (built_in) block size 4096	58774.56 MByte/s (±42668.24 MByte/s)	64775.93 MByte/s (±45733.35 MByte/s)	0.91
Memset speed - (built_in) block size 1048576	16256.91 MByte/s (±13359.03 MByte/s)	15762.33 MByte/s (±12632.18 MByte/s)	1.03
Memset speed - (built_in) block size 16777216	11807.47 MByte/s (±9608.00 MByte/s)	11893.62 MByte/s (±9629.31 MByte/s)	0.99
Memcpy speed - (rust) block size 4096	55149.65 MByte/s (±40584.91 MByte/s)	54010.26 MByte/s (±39296.64 MByte/s)	1.02
Memcpy speed - (rust) block size 1048576	15401.52 MByte/s (±12395.32 MByte/s)	15677.56 MByte/s (±12448.09 MByte/s)	0.98
Memcpy speed - (rust) block size 16777216	11202.90 MByte/s (±9045.11 MByte/s)	11260.20 MByte/s (±9144.57 MByte/s)	0.99
Memset speed - (rust) block size 4096	55432.20 MByte/s (±40781.79 MByte/s)	54756.81 MByte/s (±39887.77 MByte/s)	1.01
Memset speed - (rust) block size 1048576	15963.23 MByte/s (±12882.14 MByte/s)	16095.43 MByte/s (±12723.66 MByte/s)	0.99
Memset speed - (rust) block size 16777216	11636.27 MByte/s (±9425.52 MByte/s)	11569.34 MByte/s (±9343.13 MByte/s)	1.01
alloc_benchmarks Build Time	266.12 s	266.80 s	1.00
alloc_benchmarks File Size	0.96 MB	0.97 MB	0.99
Allocations - Allocation success	100.00 %	100.00 %	1
Allocations - Deallocation success	100.00 %	100.00 %	1
Allocations - Pre-fail Allocations	100.00 %	100.00 %	1
Allocations - Average Allocation time	11976.50 Ticks (±457.50 Ticks)	11934.55 Ticks (±589.33 Ticks)	1.00
Allocations - Average Allocation time (no fail)	11976.50 Ticks (±457.50 Ticks)	11934.55 Ticks (±589.33 Ticks)	1.00
Allocations - Average Deallocation time	2277.23 Ticks (±334.35 Ticks)	2238.08 Ticks (±305.33 Ticks)	1.02
mutex_benchmark Build Time	266.01 s	265.84 s	1.00
mutex_benchmark File Size	1.01 MB	1.02 MB	0.99
Mutex Stress Test Average Time per Iteration - 1 Threads	33.36 ns (±2.53 ns)	33.24 ns (±2.55 ns)	1.00
Mutex Stress Test Average Time per Iteration - 2 Threads	28.52 ns (±2.71 ns)	27.66 ns (±3.09 ns)	1.03

This comment was automatically generated by workflow using github-action-benchmark.

[Gelbpunkt]

🤔 I think I have made this exact change before, but apparently I never PR'ed it. This makes sense.