CompactDecimal and ScientificDecimal

[eggrobin]

Getting back to #1267, where we had decided to add these types.

I have a couple of questions here.

1. Is this is the right place for these structs? Edit: no it wasn’t, CompactDecimal and ScientificDecimal #2847 (comment).
2. ScientificDecimal is a lot like FixedDecimal; it has the nice property all of its values can be formatted (e.g., 0012.300e+003 would unambiguously be 0012,300×10⁺⁰⁰³ in fr-FR, which, while silly-looking, is well-defined).
   CompactDecimal is much messier; 1c4 cannot be formatted in en-US, and arguably neither can 1000c6 because of the way CLDR compact decimal data is structured.
   I think it nevertheless makes sense to use the source number as our input here, because it allows users to have their own rounding logic (something for which there is a use case, and which is very tricky to do with ICU4C; when I needed that, I resorted to counting digits in the formatted output!). However this is not directly usable without also providing the locale-dependent mapping from the magnitude of the source number to the power of ten used in compact notation (in en-US, something like 3↦3, 4↦3, 5↦3, 6↦6, etc.).
   I would like to confirm that are aware of that, and that we still want to use CompactDecimal as our source format.

[CLAassistant]

All committers have signed the CLA.

[sffc]

Great start!

[sffc]

Please update missing_apis.txt:

+fixed_decimal::CompactDecimal#Struct
+fixed_decimal::CompactDecimal::from_str#FnInStruct
+fixed_decimal::CompactDecimal::write_to#FnInStruct
+fixed_decimal::FixedInteger#Struct
+fixed_decimal::FixedInteger::from_str#FnInStruct
+fixed_decimal::FixedInteger::write_to#FnInStruct
+fixed_decimal::ScientificDecimal#Struct
+fixed_decimal::ScientificDecimal::from#FnInStruct
+fixed_decimal::ScientificDecimal::from_str#FnInStruct
+fixed_decimal::ScientificDecimal::write_to#FnInStruct

Make sure you merge in main first.