fix(stdlib): Only compute the garbage embedded_curve_result result if we know we will need it

[aakoshh]

Description

Problem*

Resolves #9923

Summary*

Changes embedded_curve_add in the stdlib to only compute the result on the points which are artificially guaranteed to be different if we know that we are doubling the input. Otherwise this is an operation that is guaranteed to fail as x+1 for abscissae is not on the curve, and then the circuit only succeeds when DIE removes the unused instruction.

Brillig never failed because it doesn't use this method, going straight to embedded_curve_add_unsafe.

This will be part of integration tests once #9912 is merged, where any Err observed in interpret will have caused the command to fail.

Additional Context

Here's the SSA before and after it "heals". As we can see v7-v20 are unused and eliminated, but embedded_curve_add(v7, v2, ... will fail because v7 is add v1, Field 1 which is not on the curve, while embedded_curve_add(v1, v2, ... works.

After Mem2Reg (step 9):
acir(inline) predicate_pure fn main f0 {
  b0(v0: Field, v1: Field, v2: Field):
    v7 = add v1, Field 1
    v11 = call embedded_curve_add(v7, v2, u1 0, v1, v2, u1 0, u1 1) -> [(Field, Field, u1); 1]
    v13 = array_get v11, index u32 0 -> Field
    v15 = array_get v11, index u32 1 -> Field
    v17 = array_get v11, index u32 2 -> u1
    v18 = allocate -> &mut Field
    v19 = allocate -> &mut Field
    v20 = allocate -> &mut u1
    v21 = call embedded_curve_add(v1, v2, u1 0, v1, v2, u1 0, u1 1) -> [(Field, Field, u1); 1]
    v22 = array_get v21, index u32 0 -> Field
    v23 = array_get v21, index u32 1 -> Field
    v24 = array_get v21, index u32 2 -> u1
    jmp b1()
  b1():
    jmp b2(v22, v23, v24)
  b2(v3: Field, v4: Field, v5: u1):
    jmp b3()
  b3():
    jmp b4()
  b4():
    return v3, v4, u1 0
}

--- Interpreter result after Mem2Reg (step 9):
Err(Failed to solve blackbox function embedded_curve_add: Point (0000000000000000000000000000000000000000000000000000000000000002, 0000000000000002cf135e7506a45d632d270d45f1181294833fc48d823f272c) is not on curve)
---
After Dead Instruction Elimination (step 10):
acir(inline) predicate_pure fn main f0 {
  b0(v0: Field, v1: Field, v2: Field):
    v8 = call embedded_curve_add(v1, v2, u1 0, v1, v2, u1 0, u1 1) -> [(Field, Field, u1); 1]
    v10 = array_get v8, index u32 0 -> Field
    v12 = array_get v8, index u32 1 -> Field
    jmp b1()
  b1():
    jmp b2(v10, v12)
  b2(v3: Field, v4: Field):
    jmp b3()
  b3():
    jmp b4()
  b4():
    return v3, v4, u1 0
}

--- Interpreter result after Dead Instruction Elimination (step 10):
Ok(Field 3078034153852398078128400807926804309327113743808504829582559963737223069694, Field -9191351987198133172789796342745422989482268917117950487758512501574271532885, u1 0)
---

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Test Suite Duration'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: 97961f7	Previous: 1ffaa3b	Ratio
test_report_zkpassport_noir-ecdsa_	3 s	2 s	1.50

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Execution Time'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: 97961f7	Previous: 1ffaa3b	Ratio
rollup-block-root-single-tx	0.003 s	0.002 s	1.50
rollup-checkpoint-merge	0.004 s	0.003 s	1.33
sha512-100-bytes	0.095 s	0.075 s	1.27

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[vezenovm]

LGTM

Some snaps need updating it looks like