Describe the bug
FIPS charts that group by the year_from derived field fail at runtime with a MongoDB aggregation pipeline error. The year_from field uses a hardcoded reference to the Common Criteria date field not_valid_before, which does not exist in FIPS documents. FIPS stores its certification date in web_data.date_validation. When the pipeline attempts to extract a year from the missing field, MongoDB throws a $convert error.

To Reproduce
Steps to reproduce the behavior:

1. Start the Flask dashboard application
2. Log in as a user with the dashboard role
3. Navigate to the FIPS dashboard
4. Open any chart that uses year_from as its X-axis grouping field (e.g., FIPS Standard Migration Over Time)
5. The chart fails to render, and the server log shows: Field 'year_from' not found in data

Expected behavior
FIPS charts using year_from should extract the year from web_data.date_validation (the FIPS date field) and render correctly, the same way CC charts extract the year from not_valid_before.

Desktop (please complete the following information):

• OS: Windows
• Version of sec-certs: current page branch
• Did you use dockerized sec-certs or Bare Python? Bare Python

Additional context
The root cause is that DERIVED_FIELD_EXPRESSIONS in query_builder.py is a flat dictionary with a single DerivedFieldDefinition per derived field. There is no mechanism to provide collection-specific aggregation expressions. The fix restructures it into a nested dictionary keyed by CollectionName, with a resolve_derived_field(field, collection_name) helper that selects the correct variant per collection (or falls back to a collection-agnostic default).