SIP310 Implement H2O mass-balance checking

[ANAMASGARD]

FIXES #310

Summary

Implements water mass-balance checking following the same framework as PR #248 (carbon/nitrogen):

• Tracks soilWater + snow each timestep (pre-update → post-update → post-clamp)
• Compares pool change against external water fluxes (rain, snowfall, irrigation, transpiration, evaporation, sublimation, drainage, fast flow)
• Logs a warning when deltaWater exceeds tolerance (EPS)
• Adds bcdeltaC, bcdeltaN, and bcdeltaW columns to sipnet.out
• Extends testBalance with checkWater() and updates smoke baselines

Design notes

• No leaf-water pool — ctx.leafWater only caps immedEvap; excess interception goes to soil via netRain
• snowMelt excluded — internal transfer (snow → soil), not external I/O
• eventEvap on both sides — gross irrigation enters the system (input) and the evaporated portion leaves immediately (output); net soil addition is eventSoilWater

Test plan

• make testbuild — builds successfully
• Water balance closes: bcdeltaW = 0.00000 in smoke output; zero "Water balance check failed" warnings
• checkWater() passes in all testBalance subtests
• make smoke — passes after baseline update

[Alomir]

testBalance overall still fails due to a pre-existing carbon balance delta (deltaC ≈ 610 g C/m²), unrelated to this PR. Water checks pass.

@ANAMASGARD Can you say more about this? I don't see the failure.

[ANAMASGARD]

@Alomir sorry for the confusion, and also sorry my previous reply missed your actual question entirely (I answered the wrong thing).

You're right that you don't see the failure - because it wasn't a real pre-existing issue. That note was based on a testBalance failure I saw only in my local environment before syncing with master. After merging master, it disappeared completely and all tests pass cleanly in CI. I can't reproduce it anymore either.

It was likely a stale local build artifact rather than an actual code issue. I've removed that checkbox from the PR description to avoid further confusion.