RFC: 64-bit integer type proposal improvements

[vegorov-rbx]

This PR adds changes to the RFC following additional user feedback and early usage feedback.
It also fixes more wording issues and adds clarity to some changes that were not mentioned.

• goes over how negative integer literals are supported
• min/max are variadic to match the math library they mirror
• band/bor/box/btest are variadic to match the bit32 library they mirror
• gt/ge/ugt/uge comparisons to not worry about transforming the order to fit into the small set of operations
• explicitly mention the change to tostring, type and typeof
• mention support in rawequal; similar to previous base library functions, it is expected to work on any
• remove integer.tostring since it duplicates tostring
• add types.integer constant for use in type functions
• math.mininteger and math.maxinteger moved to integer.minsigned and integer.maxsigned

[AsynchronousAI]

Hi, I don’t know if this is the place to mention but a popcount function would also be helpful.

Not a requirement tho because it can still be manually implemented to be fast with countlz and countrz and a pop_lsb but having it be O(1) with CPU instructions would be extremely useful.

[vegorov-rbx]

Popcnt does make sense and it was considered.
The reason it was excluded is that it makes sense for bit32 library as well, so it could be a separate proposal to add to both.

[AsynchronousAI]

Popcnt does make sense and it was considered. The reason it was excluded is that it makes sense for bit32 library as well, so it could be a separate proposal to add to both.

One more thing, will a buffer.writei64 function be added?

[ahae-roblox]

Do we want to support string.pack()/string.unpack()?

[vegorov-rbx]

Added a section about it, it doesn't seem backwards compatible to accept integers in those functions.

[MagmaBurnsV]

Just to clarify, what would -0xffffffffffffffffi be? Because depending on how that is parsed, it could either be 1 or -9223372036854775807.

Inversely, is -0x8000000000000000i invalid from overflow, or does it wrap back around to 0x8000000000000000i due to two's complement?

[vegorov-rbx]

-0xffffffffffffffffi is 1i, -0x8000000000000000i is 0x8000000000000000i.

[vrn-sn]

Wanted to mention that as part of this PR, maybe it's worth also amending this section of the math constants RFC (math-constants.md):

Integer types are not supported in Luau, so math.maxinteger and math.mininteger wouldn’t make much sense, but adding other commonly used constants would be an easy value-add for users of the language.

It's worth keeping up-to-date since Luau does now support integers, along with support for the integer library constants instead.

Maybe: "Luau supports similar integer constants through its integer library, but other commonly used math constants are an easy value-add for users of the language."

[vegorov-rbx]

Maybe: "Luau supports similar integer constants through its integer library, but other commonly used math constants are an easy value-add for users of the language."

Applied.