[perf] Use a flat token stream representation when lexing

[Kobzol]

Opening for a perf experiment, nothing to see here.

[Kobzol]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 8374c7c (8374c7ca6b40db24e90f1c62568ddd867a1d48ce)
Base parent: ab4bbeb (ab4bbeb07dc7dc2d30e9f83d3e0326df981e0ee2)

[rust-timer]

Finished benchmarking commit (8374c7c): comparison URL.

Overall result: ❌✅ regressions and improvements - please read:

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

Next, please: If you can, justify the regressions found in this try perf run in writing along with @rustbot label: +perf-regression-triaged. If not, fix the regressions and do another perf run. Neutral or positive results will clear the label automatically.

@bors rollup=never rustc-perf
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	0.4%	[0.1%, 1.2%]	73
Regressions ❌ (secondary)	0.6%	[0.1%, 1.4%]	59
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.8%	[-2.4%, -1.3%]	13
All ❌✅ (primary)	0.4%	[0.1%, 1.2%]	73

Max RSS (memory usage)

Results (secondary -6.6%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.9%	[1.8%, 8.5%]	14
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-13.6%	[-22.8%, -2.3%]	21
All ❌✅ (primary)	-	-	0

Cycles

Results (primary 1.1%, secondary -7.3%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	2.6%	[2.4%, 3.0%]	3
Regressions ❌ (secondary)	3.5%	[2.1%, 5.2%]	8
Improvements ✅ (primary)	-3.4%	[-3.4%, -3.4%]	1
Improvements ✅ (secondary)	-13.6%	[-19.9%, -4.9%]	14
All ❌✅ (primary)	1.1%	[-3.4%, 3.0%]	4

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 489.393s -> 489.02s (-0.08%)
Artifact size: 389.94 MiB -> 389.88 MiB (-0.02%)

[Kobzol]

Seems like doing the flat representation in the lexer doesn't help much when we still have to convert it back to all the allocations required for the parser. The whole thing would have to be flattened to have the desired perf. effect.