fix: BigInt exponentiation transpiler error (@dfinity/candid)

[rvanasa]

Description

Fixes a bug encountered in browser environments such as Motoko Playground where BigInt(a) ** BigInt(b) transpiles to Math.pow(BigInt(a), BigInt(b)) for compatibility with older browsers (resulting in a runtime error). This PR adds iexp2 and ilog2 implementations which account for BigInt values and provide clearer error messages for invalid input.

@chenyan-dfinity

How Has This Been Tested?

• Full repository test suite
• Additional test cases for the iexp2 and ilog2 functions
• Confirmed to fix the issue in Motoko Playground

Checklist:

• My changes follow the guidelines in CONTRIBUTING.md.
• The title of this PR complies with Conventional Commits.
• I have edited the CHANGELOG accordingly. (N/A, worked as expected in 0.11.x)
• I have made corresponding changes to the documentation. (N/A)

[chenyan-dfinity]

LGTM

[krpeacock]

I see - this works because we don't have any mandatory exponentiation other than 2, and don't need to support generalizing inside of the lib. Clever!