handle: duplicate task names

[metaist]

We need some way of handling ambiguous / duplicate task names. When we implement #8 we run the risk of having the same name in multiple sections.

[tool.ds.scripts]
a = "echo 'ds:a'"
b = "echo 'ds:b'"

[tool.pdm.scripts]
a = "echo 'pdm:a'"
c = "echo 'pdm:c'"

What should the result be?

[metaist]

Idea 1: later names (per resolution order) need to specify a prefix. So the result would be:

{
    "a": "echo 'ds:a'",
    "b": "echo 'ds:b'",
    "pdm:a": "echo 'pdm:a'",
    "c": "echo 'pdm:c'"
}

This is the simplest to implement, but feels weird. You'd have to know the resolution order to know which thing will get prefixed.

[metaist]

Idea 2: when adding a name that would cause a conflict, both names get prefixed.

{
    "a": "echo 'ds:a'",
    "ds:b": "echo 'ds:b'",
    "pdm:a": "echo 'pdm:a'",
    "c": "echo 'pdm:c'"
}

This feels fairer: you have two tasks with the same name, please be more precise (and the previous task b is no invalid). We could offer hints about what you might have meant.

[metaist]

Idea 3: if multiple tools have configuration in the same file, all names are prefixed.

{
    "ds:a": "echo 'ds:a'",
    "ds:b": "echo 'ds:b'",
    "pdm:a": "echo 'pdm:a'",
    "pdm:c": "echo 'pdm:c'"
}

The main advantage is to discourage having multiple tools configure scripts and to name the tasks consistently. Disadvantage is that you have to type more (although you could just do pdm b if you were really using pdm).

[metaist]

Idea 4: Follow a resolution order and only parse the first compatible section.

This encourages only putting the configuration in one place and reduces the surprise for the task names.