BUG: RuntimeWarning in snow.categorize and outliers.zscore for edge case inputs

[Omesh37]

Bug description

Two functions emit RuntimeWarning for valid but edge-case inputs:

1. pvanalytics/features/snow.py:229

RuntimeWarning: invalid value encountered in divide when modeled_voltage_with_snow_copy contains zeros.

The function already uses np.errstate(divide='ignore') but does not suppress invalid which fires on 0/0.

2. pvanalytics/quality/outliers.py:78

RuntimeWarning: Precision loss occurred in moment calculation when all input values are identical (constant data). This comes from scipy.stats.zscore internally.

To reproduce

# snow.py warning
pytest pvanalytics/tests/features/test_snow.py::test_categorize -W error::RuntimeWarning

# outliers.py warning  
pytest pvanalytics/tests/quality/test_outliers.py::test_zscore_all_same -W error::RuntimeWarning

Fix

# snow.py — added 'invalid' to existing errstate
with np.errstate(divide='ignore', invalid='ignore'):
    vmp_ratio = measured_voltage / modeled_voltage_with_snow_copy

# outliers.py — wraped zscore call
with np.errstate(invalid='ignore'):
    is_outlier.loc[~nan_mask] = abs(stats.zscore(data[~nan_mask])) > zmax