A real-world ABC9 flow hit residual combinational loops after SCC breaking, tripping the prep_xaiger loop assertion.
Keep the existing topological assertions in place (prep_xaiger and reintegrate still assert no_loops).
To handle residual non-box loops, add a targeted fallback in prep_xaiger: when loops remain after normal SCC breaking, insert additional $__ABC9_SCC_BREAKER cuts on non-box loop cells, rebuild toposort, and then re-check the existing assertion.
Also keep pre-ABC9 SCC tagging on all cell types (scc -all_cell_types) and add a regression test (tests/techmap/abc9-nonbox-loop-with-box.ys).
The main part is converting ice40_dsp to recognize the new FF types
created in opt_dff instead of trying to recognize the mux patterns on
its own.
The fsm call has been moved upwards because the passes cannot deal with
$dffe/$sdff*, and other optimizations don't help it much anyway.
