Normalize duration function about date-fns HOT 1 OPEN

Other date and time library like luxon start from the lowest to the greatest unit.
There is no standard for normalising with some AI knowledge, and research I found two option:

• Conversion to Milliseconds: While straightforward, this method might suffer from precision loss due to floating-point arithmetic. It can be efficient for large duration ranges but may not be ideal for high-precision needs.
• Unit-by-Unit Cut and Distribute: This approach iterates through units from smallest to largest, performing rounding/truncation and carrying over excess to the next unit. It can be more accurate for shorter durations and finer units but may be less efficient for very large durations.

Other considerations can be:

• Granularity Level:

If high precision is required, normalization to smaller time units (seconds, milliseconds) may be necessary.
However, this can increase the complexity and computational burden of the process.

• Contextual Understanding:

Normalization must be consistent with the meaning and usage of the duration in the specific context.
For example, normalizing a flight duration may differ from normalizing an industrial process duration.

• Computational Efficiency:

Normalization can be a complex operation, especially for long periods of time.
It is important to evaluate the computational impact of normalization and choose an efficient method.

from date-fns.