Openfhe: get mulDepth from LWE type

[ZenithalHourlyRate]

Related to #1145. This is a step towards a generic configuring crypto context pass, as the Lattigo backend also needs one.

Since we have RNS lowering for secret-to-, we can get mulDepth from LWE type.

Still marked as draft as the CKKS RNS lowering has not landed. All tests working when merged with #1176

Several changes

• hasMulOp becomes hasRelinOp as we actually does not emit openfhe.mul (implicit relin)
• the configure_crypto_context_complex.mlir removed scf.if and used arith.add instead, as
  • we can not lower correctly through a plaintext scf.if for now
  • for oblivious if, arith.select also does not have a lowering

[ZenithalHourlyRate]

Rebased and Marked as ready since #1176 merged.

[j2kun]

Just for confirmation: this is 2 now because of the inclusion of the optimal relin insertion pass (or rather, the maximum constraint within that pass that prevents going to depth 3).

[ZenithalHourlyRate]

It is due to operation-balancer.

Without them we get mulDepth 3 but with them (--secretize --wrap-generic --cse --canonicalize --operation-balancer --secret-insert-mgmt-bgv --secret-distribute-generic --secret-to-bgv=poly-mod-degree=32 --openfhe-configure-crypto-context=entry-function=complex_func) we get mulDepth 2.

[ZenithalHourlyRate]

I put the detailed flag in the test so we still get mulDepth = 3. Otherwise it is too confusing.

[ZenithalHourlyRate]

Oh this one should wait until #1190

[ZenithalHourlyRate]

Rebased and ready now.

[AlexanderViand-Intel]

LGTM👍, just a few questions/thoughts (that shouldn't stand in the way of merging this)