Add initValue1DefaultBranch test case to btree remove module - #1490

Approving. Reviewed at fb07366. Three lines, and they do exactly what the title says. npx tsc clean, 2357 pass / 0 fail — one more than main's 2356, which is the new case.

The test isn't vacuous

A case under proof.throw passes when its body throws, so the thing worth checking is whether it would still pass if the branch it targets stopped throwing. I mutated initValue1's default arm to return instead:

-            default: { throw 'invalid node' }
+            default: { return [[...n0, v1]] }

before:  proof.throw.initValue1DefaultBranch(): ok  # EXPECTED TO THROW   → 2357 pass, 0 fail
after:   proof.throw.initValue1DefaultBranch(): error                     → 2356 pass, 1 fail

So it genuinely pins the guard. The inputs are right for it too: a === null with n0.length === 3 is the one shape that reaches default, since case 1 and case 2 take the other arms and a non-null a skips the switch entirely.

Not asserting the thrown payload is correct here, per §3's note that whether it threw is normally the part of the contract that matters.

The comment above the proof is the part I'd single out — it explains that these arms are unreachable through the public remove API because the sibling is always a Branch in any reachable tree, which is exactly the question a reader has when they see a guard being exercised directly rather than through the API.

A natural follow-up, not a request

All four merge helpers carry the same guard:

This takes the set from 1/4 to 2/4, and the two remaining ones have no proof case anywhere in the repo. The comment already describes all four as a group, so closing the other two would be the same two lines each. Out of scope for a PR that says it covers initValue1 — just flagging it so the gap is visible rather than implied to be closed.