[codex] Use channel energy for hit smearing

[lobis]

Summary

• Compute the smearing resolution from the energy in the selected channel type instead of the total detector-hit energy.
• Keep the previous behavior for applying one common stochastic gain to the selected TPC or VETO hits.
• Leave non-selected hit types unchanged and avoid drawing a smearing factor when the selected channel has zero energy.

Why

TRestDetectorHitsSmearingProcess can be configured with channelType=tpc or channelType=veto, but the resolution was derived from fInputEvent->GetTotalEnergy(). For mixed events containing both TPC and VETO hits, a large energy deposit in the other channel could bias the selected-channel resolution. This is especially visible in cosmic-background simulations where a small TPC deposit may coexist with a large veto deposit.

Validation

• Ran git diff --check on the patch.
• Built the RestDetector target on the NAF REST build tree successfully.

[lobis]

Follow-up verification on 2026-05-25: this patch is still not present on master and remains isolated to src/TRestDetectorHitsSmearingProcess.cxx.

I also rechecked the failing CI job. The only failing check is restG4-validation / Example 12: Generators, which exits from ValidateCosmicGenerator.C with wrong cosmic flux: 0.0172229. That appears tied to the Guan cosmic-muon normalization update in geant4lib, not to this detector-hit smearing patch. The detector-specific validation/build checks are green, and the patched RestDetector target was rebuilt and installed successfully in the NAF REST stack.