Better document the behavior of nested function contracts

[yannham]

When several function contracts are applied, the order of contract application is reversed for arguments, cf #1358 (comment). This can happen in particular with our "differential contract" approach (type annotation + contract annotation), see #1358. This isn't a bug or any specific feature of Nickel, but just the logical conclusion of how higher-order function contracts work. It could be however useful to write a short explicit explanation in the manual.