Improve clarity of documentation for focal weights

[ailich]

I've included some changes that I think would clarify w for focal. Some changes to the documentation include:

• preferentially suggesting the use NA over 0 in w as this is safer since it masks those cells out of calculations rather than setting them to 0 which may affect calculations depending on the function.
• Describes what the weights do in calculations and removes restriction that they must be between 0 and 1.
• Slight edit to special case of mean stating that the weighted mean will be calculated rather than just saying 0's function like NA's.

[rhijmans]

Thank you very much!