Fix MPO_new assembly cutoff and add H1 accuracy regression test

[srwhite59]

Problem

MPO_new can lose one-body accuracy (~1e-7) on a minimal reproducer where reference MPO(...; cutoff=0.0) is ~1e-11.

Root cause

During tensor assembly in src/MPOConstruction.jl, entries are filtered with a hard-coded tol=1e- 10:

• my_ITensor(...; tol=1e-10, ...)
• itensor(...; tol=1e-10, ...)

This creates an additional implicit truncation independent of the user-facing MPO_new(...; tol=...).

Change

Replaced the fixed assembly cutoff with exact assembly (tol=0.0) in both call sites.

  - tol=1e-10
  + tol=0.0

Validation

Added a minimal accuracy loss test case in test/

On the reproducer case:

• before: MPO_new max|Δ| = 3.132e-07
• after: MPO_new max|Δ| = 2.910e-11
• reference MPO(...; cutoff=0.0): 4.093e-11

Why this is correct

MPO_new approximation should be controlled by its decomposition logic/tolerance settings. A fixed
assembly-time drop introduces uncontrolled representation error and can dominate one-body
expectation values.

[codecov]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[corbett5]

Looks good to me. I appreciate the inclusion of the test as well.

I don't even know if I had any justification for choosing tol=1e-10 other than it sounded like a good idea.