fix: Handle aggregate expressions in chart sorting

[sentry]

This PR addresses the ProgrammingError: (psycopg2.errors.UndefinedColumn) column "SUM(silt_achieved)" does not exist that occurred when attempting to sort charts by aggregate expressions (e.g., SUM(silt_achieved)).

Root Cause:
The apply_chart_sorting() function in ddpui/core/charts/charts_service.py was incorrectly treating aggregate expressions provided in the sort configuration as raw column names. When SQLAlchemy processed these, it would quote them as literal identifiers (e.g., "SUM(silt_achieved)"), which do not exist as actual columns in the database, leading to the UndefinedColumn error.

Solution:

1. Enhanced Parsing: The apply_chart_sorting() function now includes logic to detect and parse aggregate-expression-style sort column strings (e.g., SUM(column_name)) using a regular expression.
2. Metric Matching: After parsing, it first attempts to match the extracted aggregation and column name against the payload.metrics to find a corresponding metric. If a match is found, the metric's alias is used for sorting.
3. SQLAlchemy Expression Generation: If no direct metric match is found, the function now dynamically constructs the appropriate SQLAlchemy func expression (e.g., func.sum(column('column_name'))) for the ORDER BY clause. These expressions are directly appended to query_builder.order_by_clauses, bypassing the order_cols_by method's default behavior of quoting string column names.

This ensures that aggregate expressions in sort configurations are correctly translated into valid SQL, resolving the UndefinedColumn error.

Fixes DALGO-BACKEND-29P