Subtract dynbal baselines from begwb and endwb

In a branch that I'll soon bring to master, I'm subtracting baselines from the calculation of total column water and heat when computing budgets before and after dynamic landunit adjustments. However, I am currently NOT subtracting these baselines when computing begwb and endwb for the sake of water balance checks - and also for the sake of TWS, since TWS is computed based on endwb.

It would be better if these different calculations of column-level water were consistent, always subtracting these dynbal baselines (`dynbal_baseline_liq_col` and `dynbal_baseline_ice_col`). However, this changes answers by more than we'd like right now, presumably due to methane's dependence on gridcell TWS (see #658 - this subtraction of baselines should only change answers for glacier columns, but this can affect non-glacier columns due to the dependence on gridcell average TWS). We want to avoid these answer changes on the release branch, and if possible also on master (partly to keep master consistent with the release branch in this respect for now, and partly because master still may be used in some runs that extend cmip6 runs, adding water isotopes).

Once it's okay to change answers on master - and/or once #658 is resolved so that methane no longer depends on gridcell TWS - we should clean this up, _always_ subtracting these dynbal baselines in the calculations of total column water.