Some kinds of maps, such as alists and plists, can contain duplicate
keys.  These duplicates are effectively ignored in functions like
`map-elt`, but are not ignored in functions like `map-do` or `map-length`.

To me, it would make more sense if these functions only considered the
valid entries in the map.  For example,

    (map-pairs '(a 1 b 2 a 3))

currently returns '((a . 1) (b . 2) (a . 3)), even though '(a . 3) is
meant to be ignored in actual usage, as it is preceded by an earlier
entry with the same key.  This is a misleading behavior.

I do not know whether the current behavior is desirable.  Is it?

Please consider changing the library so that the duplicate,
meant-to-be-ignored entries are actually ignored for functions that
operate on the entire map.

I have attached an example diff.

Thank you.