[numeric.limits.members] Correct footnotes and convert to notes and examples

[eisenwave]

There are many major issues with [numeric.limits.members] right now:

• Many of the footnotes say something like "Equivalent to FLT_RADIX". This doesn't account for double being a decimal floating-point type (which is permitted I believe), and doesn't account for any extended floating-point types that are not binary.
• Many footnotes shouldn't be footnotes. To my understanding, footnotes should only be used for references to external documents, not for immediate clarifications or examples.
• Many footnotes are exhaustive when they shouldn't be, by only mentioning the FLT, DBL, and LDBL macro. However, with extended floating-point types, the value of the member can obviously be soemthing other than those three, so there needs to be a etc. somewhere.
• There is a footnote attached to the description of radix for integers, accounting for representations like BCD. Seeing that since C++20, the representation of integers is always two's complement, this footnote no longer makes any sense.

This PR cleans up all those problems.