Eliminate need for multiple calls to nondeterministic millis() when getting individual day/time components

[LeonPoon]

Also eliminated some dup calc code between getHours/Minutes/Seconds and getFormattedTime.

Fully backwards compatible.

[CLAassistant]

All committers have signed the CLA.

[LeonPoon]

Would appreciate feedback, or a merge please.

I encountered one instance when the getHours and getMinutes don't provide consistent results, because of each of these make calls to millis() on their own, which might not return the same value. It was 8:59++ and getHours() returned 8 but getMinutes returned 0 and my datalogger recorded time as 8:00 when it was actually closer to 9:00.

With this patch, allows me to provide same millis input to gethours/minutes rather than having these get functions call their own millis().

thanks.