[Pyhroff]

Fixes #1818.

Summary

• In Add.decompose_from_registers, when �_dtype.bitsize < b_dtype.bitsize (m < n), the left building block elides
• 1 - m CNOTs (treating missing high bits of � as |0>), the top-bit CNOT (len(input_bits) == len(output_bits)) is omitted, and the right building block elides 2 * (n - 1 - m) CNOTs.
• Previously, Add.build_call_graph unconditionally returned
  _cnot = (n - 2) * 6 + 3, causing gate.t_complexity() == gate.decompose_bloq().t_complexity() and �ssert_equivalent_bloq_counts to fail whenever �_dtype.bitsize < b_dtype.bitsize (or when �_dtype.bitsize == 1).
• Updated Add.build_call_graph to account for m = self.a_dtype.bitsize vs
  = self.b_dtype.bitsize (yielding (n - 2) * 3 + 2 + 3 * m - 3 CNOTs when m < n, (n - 2) * 6 + 3 when m == n, and 1 CNOT when
  == 1) and added parametrized regression tests in �ddition_test.py.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.