[Payables Agent] Enhance historical data loading with error handling and telemetry

[dayland]

This pull request introduces enhanced error handling and telemetry logging to the historical data loading process in the E-Doc. Historical Matching codeunit. The main improvements are focused on making the data load process more robust and observable, especially in error scenarios and performance tracking.

Error handling and telemetry improvements:

• Added a new [TryFunction] procedure TryLoadHistoricalData to wrap the historical data loading, enabling graceful handling of load failures.
• On failure to load historical data, the code now logs a detailed error event using FeatureTelemetry.LogError, including the e-document system ID and error message for better diagnostics.

Performance and usage tracking:

• Enhanced the LoadHistoricalDataIntoTempTable procedure to measure and log the time taken, number of records loaded, and whether the record limit was reached, using FeatureTelemetry.LogUsage and a telemetry dimensions dictionary.

Supporting changes:

• Added new label constants for error messages and event tokens (HistoricalDataLoadFailedErr, AIHistoricalDataLoadEventTok) to improve error reporting and event tracking.
• Declared additional variables for telemetry, error handling, and data tracking (FeatureTelemetry, EDocSystemId, ErrorMessage, TelemetryDimensions) to support the new functionality.

Fixes AB#604259

[Groenbech96]

Just thinking a bit on all the AI tools we have, maybe we could move this to reused across all tools for consistent telemetry? Wdyt

[dayland]

At the moment this is the only place we load large amounts of historical data, and I think this pattern will go away when we move to a model-first approach. So, I don't think refactoring this now to make generic is needed IMO.