Keep a placed switch its own input when a new input is placed

[rollrat]

Bottom of a stack: this fix, then the e-graph netlist search, then OR nets for wide NORs.

Why

A construction step that placed a new input sometimes failed as infeasible within 0.1 s, for every window, and construction then restarted with a new seed. This happened in three of five seeds of one full-adder netlist and, earlier, at the carry tile's carry entry. It wasted whole construction attempts and hid which netlists can actually be placed.

What

• Cause: the new input's candidate sites covered the whole window, including the overlap slice, where a switch placed in an earlier step is kept as a fixed cell. dsl.rs gave a fixed switch to the first input with a site there. The kept switch became the new input's, so the earlier input lost its only site, and the frozen gates it drove contradicted their given signals.
• Fix: new inputs no longer get sites where a switch is already placed. A fixed switch belongs to the input whose only site it is when several inputs could claim it (an error if that is still ambiguous).
• Diagnosis hook: with EXACT_DIAGNOSE_INFEASIBLE, an infeasible step is re-solved with given signals, each observation, each single-site input, and frozen air dropped in turn, and the placed inputs' surroundings are logged. That is how the cause was found: only freeing the kept switch made the step solvable.

The same seed now places that step in 9 s.

Test plan

• cargo test --release --lib exact::: all pass.
• The failing seed of CIRCUIT=egraph-full-adder EGRAPH_DEPTH=3 now places the step.

🤖 Generated with Claude Code

[rollrat]

Opened by mistake as a separate stack; the work stays in #61 on top of #59.