Try including the generated grammar in-tree

[jneem]

This is an attempt at #2123, by checking in a version of the generated grammar. The checked in file is 22MB (!), but I checked that the compressed cargo package size is more reasonable (1.2 MB). crates.io has a limit of 10MB per crate.

Stepping back, the main goal here is to avoid making users of the published nickel-lang-core generate the grammar when building. But we also need to make sure that it isn't too painful to modify the grammar. I think there are two basic approaches

1. check in the generated grammar, and require some workarounds when modifying it
2. generate the grammar at publish time, so it's in the published crate but not in the git repo.

Option 2 seems nicer in principle, but I'm worried that because publishing is infrequent and mostly automated, it will be prone to undetected breakage. So this PR does option 1, but I think the mechanism in the build script would work with either one; option 2 would just need some more work on the release script.

[github-actions]

Bencher Report

Branch	check-in-the-grammar
Testbed	ubuntu-latest

Click to view all benchmark results

Benchmark	Latency	microseconds (µs)
fibonacci 10	📈 view plot 🚷 view threshold	497.16
foldl arrays 50	📈 view plot 🚷 view threshold	1,764.00
foldl arrays 500	📈 view plot 🚷 view threshold	6,859.80
foldr strings 50	📈 view plot 🚷 view threshold	7,278.70
foldr strings 500	📈 view plot 🚷 view threshold	63,406.00
generate normal 250	📈 view plot 🚷 view threshold	44,443.00
generate normal 50	📈 view plot 🚷 view threshold	2,089.60
generate normal unchecked 1000	📈 view plot 🚷 view threshold	3,273.80
generate normal unchecked 200	📈 view plot 🚷 view threshold	767.60
pidigits 100	📈 view plot 🚷 view threshold	3,222.40
pipe normal 20	📈 view plot 🚷 view threshold	1,533.20
pipe normal 200	📈 view plot 🚷 view threshold	10,050.00
product 30	📈 view plot 🚷 view threshold	859.17
scalar 10	📈 view plot 🚷 view threshold	1,555.00
sum 30	📈 view plot 🚷 view threshold	853.13

🐰 View full continuous benchmarking report in Bencher

[yannham]

I haven't reviewed, but I want to point out one of LALRPOP co-maintainer's interesting remark: https://github.com/orgs/lalrpop/discussions/1031#discussioncomment-11908105. There's a checksum of the lalrpop grammar file included somewhere, so maybe we can just include a check in the build script so that if the file has changed we rebuild it automatically?