Test command files were rewritten using:
```
for f in tests/**/*.cmd; do jq ".done?.[].Autoroute[0] |= .selectors" "$f" | sponge "$f"; done
```
(compatible with bash/zsh/...)
This was making the CLI build and tests fail because my second previous
commit was botched.
For future reference, I fixed the *.cmd files by running the following
in Fish shell:
```
for f in tests/**.cmd
jq ".done?.[].Autoroute[1].router_options.routed_band_width |= 100.0" $f | sponge $f
end
```
It's about pushing through a constrained space, not about squeezing
per se. I'm not using the word "push" because this term will probably be
restricted to a mechanism that is very different internally.
Disambiguate between node indices and (teir inner) petgraph indices.
Rename "band index" to "band first seg index", as we identify bands by
teir first segs. Rename "trianvertex index" and "navvertex index"
respectively to "trianvertex node index" and "navvertex node index", as
these point to nodes on the geometry graph and only indirectly to
vertices on the vertex graph.
The trait's contents aren't implemented yet.
"mesa" is supposed to mean the opposite of "meta". By "mesadata" I don't
mean mere content or payload: I mean data that is stored *inside* or
*under* such content or payload.
I found this meaning of "mesa" -- as opposite of "meta" -- by browsing
somewhere on the internet. Fair chance many classicists would think
poorly of this usage. But I don't care about etymology: I just need a
word to close a lexical gap.
We call these pairs "selectors".
The command file format has been thus changed. I've upgraded the command
files in tests with the following command:
jq ".done?.[].Autoroute |= {selectors: ([.pins.[] | {pin: ., layer: \"F.Cu\"}])}" $FILE | sponge $FILE
Where $FILE is the name of the upgraded command file.
Rename "triangulation vertex" to just "trianvertex".
Rename "navigation vertex" to just "navvertex".
The standalone word "vertex" was overused and in some cases confusing.