fix #8821

[ringabout]

Added a limitation for total length of range.

Choose 65535 because it generates 1 MB JS file for simple program

proc isInt32(i: int): bool =
  case i 
  of 1 .. 65536:
    return true
  else:
    return false

echo isInt32(1)

[Varriount]

I read the original issue, but it wasn't very clear - why do "extreme" ranges cause JS memory issues?

[ringabout]

It just causes compilation program crash when compiling Nim to JS. Because it generates too many labels for JS program. It's hard for me to optimize range branches. We should set a limitation in case that compilation program(NIm JS) crash.

[bung87]

Your reason to choose 65535 doesn't make sense, you will gonna explain why " 1 MB" and why "simple program"

[ringabout]

It's a temporary solution and the issue is still open(The right solution is to change tranf.nim). If users have some problems with them, it's easy to change the number. This is not backported and won't be shipped until Nim 1.6.0 (namely many months).
See #8821 for details

[narimiran]

This is not backported

I did backport it, because based on the PR title I thought it is a simple non-breaking fix. I can revert the backport if necessary.

[ringabout]

Though this is not a proper fix. But I think it doesn't cause terrible issues. If users complain this, then we could revert this change or increase or decrease threshold. Anyway it prevents compilation program from crashing when compiling to JS.