Hop to Lop instruction generation issue

[j143]

graph

            [DataOp: PRead X]
                    |
                  [TeeOp]
                 /      \
      [ReorgOp: r(r')]   [AggBinaryOp: ba(+*)]
             |                  |
     [Transpose t(X)]     [Matrix Multiply t(X)*X]

Data Flow:

PRead X → TeeOp (splits stream)
TeeOp → ReorgOp (first output, for transpose)
TeeOp → AggBinaryOp (second output, for multiply)

program plan

[CP createvar pREADX]         -- create file handle
[OOC rblk ...]                -- read blocks
[OOC tee ...]                 -- split stream into two
[OOC r' ...]                  -- transpose on first stream
[OOC ba+* ...]                -- matrix multiply (using transpose + second stream)
[CP write ...]                -- write result

graph with variable names

[PRead X] --(_mVar0)--> [TeeOp]
                |                  |
            (_mVar2)           (_mVar2_copy)
                |                  |
         [ReorgOp r(r')]       [AggBinaryOp ba(+*)]
                |                  |
            (_mVar1)           (_mVar3)
                |                  |
             [Result]

---

Hop Rewriting Works Correctly

=== BEFORE REWRITE ===
DataOp 9 has parents:
  - r(r') (Hop 10) Children: [9:PRead X]
  - ba(+*) (Hop 11) Children: [10:r(r'), 9:PRead X]

=== AFTER REWRITE ===
TeeOp 15 has parents:
  - r(r') (Hop 10) Children: [15:tee]        ✅ Now points to TeeOp
  - ba(+*) (Hop 11) Children: [10:r(r'), 15:tee]  ✅ Now points to TeeOp
 

The Hop graph is correctly rewritten - both consumers now reference the TeeOp instead of the original DataOp.

❌ But Execution Plan is Wrong

------OOC r' _mVar0.MATRIX.FP64 _mVar1.MATRIX.FP64        # ❌ Still using _mVar0
------OOC tee _mVar0.MATRIX.FP64 _mVar2.MATRIX.FP64 _mVar2_copy.MATRIX.FP64
------OOC ba+* _mVar1.MATRIX.FP64 _mVar2.MATRIX.FP64 _mVar3.MATRIX.FP64  # ❌ Should use _mVar2_copy

The problem is that the Lop→Instruction generation is not respecting the Hop graph changes.

[j143]

5f3fa08