bracket-push: rename to matching-brackets (#1501)

I rehash my prior allegations of
#693 below:

I contend that having the word `push` in the name unnecessarily biases
the solution space toward solutions that use a stack data structure (has
a *push* operation) or a *push*-down automaton, rather than other
solutions not using either of these two.

We have heard a principle that we want to name exercises by their story,
not by what they teach:
#1451 (comment)

The story here is about matching brackets, so I posit that that serves
as the name we want.

I have changed the major version number, as was done in the
retree->satellite rename, even though I do not feel strongly about
whether that should be necessary:
#1478

As we have gained experience in the rename of retree, we see that
problem-specifications is free to rename exercises at any point, without
waiting for all tracks to follow suit.
Of course, this operation should not be performed lightly since it
causes churn in the 31 (*thirty-one*) tracks implementing this exercise.

exercises/bracket-push/canonical-data.json → exercises/matching-brackets/canonical-data.json

@@ -1,6 +1,6 @@
 {
-  "exercise": "bracket-push",
-  "version": "1.5.0",
+  "exercise": "matching-brackets",
+  "version": "2.0.0",
   "cases": [
     {
       "description": "paired square brackets",