Add macros for constructing Decimals from literals

[Jules-Bertholet]

Adds three macros, d32, d64, and d128, for constructing Decimals from literals.

[benesch]

The idea of a macro for decimal literals seems sound, but I have some questions about the usability and naming:

• d32, d64, and d128 match the names of the built-in float/integer types, but are at odds with the names of the types in this package (Decimal32, Decimal64, Decimal128). I think it'd be better to follow this package's names and use decimal32, decimal64, and decimal128 as the names of the macros.
• The inclusion of the N parameter in dec is pretty rough. dec!(13, 1.0) is not super obvious syntax.

Here's an alternate proposal for consideration: we add a single macro named dec with the following usage options:

# Implicit conversion.
let d: Decimal32 = dec!(1.2);
let d: Decimal64 = dec!(1.2);
let d: Decimal128 = dec!(1.2);
let d: Decimal<13> = dec!(1.2);

# Explicit typing, mirroring the built-in integer literal suffixes.
let d = dec!(1.2_decimal32);
let d = dec!(1.2_decimal64);
let d = dec!(1.2_decimal128);
let d = dec!(1.2_decimal<13>);