Allow pattern matching on automap input and output.

Nobody has claimed this yet.

**Is your feature request related to a problem? Please describe.**
When working with Tiled, I design dungeons from a top-down perspective and have multiple levels / elevations by organizing the project into groups:
Today I must edit the layer names of the input and corresponding output layer removing the trailing index number before I start automapping the dungeon layout. And then when I want to work on an other elevation, I need to re-add the indices and start the process over.

**Describe the solution you'd like**
If the automapping could be done with pattern matching like `input_in%d` and `output_out%d` so that edits to a layer called `in2` only outputs to `out2` would be really neat.

**Describe alternatives you've considered**
I've tried reading through the documentation and the issue backlog and haven't found any workaround besides the procedure I describe.
I noticed issue #428 is sort of similar, but I think, it is somewhat orthogonal.

Start by locating Tiled's automapping implementation and the code or tests that handle input and output layer names. Trace how layer names are matched, then define completion as supporting patterns such as `in%d` and `out%d` so matching indices map to corresponding outputs without renaming layers.

Written by the indexing model from the issue text.