Describe the bug
The function predefined_conventional_loads_by_sector() in the timeseries module sets time series for different components. When ts_loads contains a time series for certain generator types or load sectors, the function scales the time series for each component differently:

• For generators, the nominal power is used for scaling.
• For loads, the annual consumption is used for scaling.

This inconsistency—scaling by power for generators but by energy for loads—is misleading and can cause unexpected behavior.

To Reproduce
Steps to reproduce the behavior:

1. Go to the timeseries module and locate predefined_conventional_loads_by_sector().
2. Pass a ts_loads input that contains time series data for both generators and loads.
3. Observe how the scaling is applied differently for generators (using nominal power) and loads (using annual consumption).
4. The output time series for loads will be unexpectedly scaled, leading to inconsistencies.

Expected behavior
The load time series should be scaled consistently. Instead of using annual consumption, the time series for loads should be scaled by p_set of the load, making it directly comparable to generator scaling.

Screenshots
(If applicable, add screenshots to help illustrate the issue.)

Additional context
This change will make the scaling logic more intuitive and prevent unexpected deviations in load time series data.