Rewrite Row_printTime() with various improvements

[Explorer09]

• Change variable types in the function from signed to unsigned.
• Change variable naming for consistency: totalMinutes, totalHours, totalDays for total values and minutes, hours, days, etc. for remainder values.
• Use unsigned long long type for years value to prevent a potential overflow.
• The time units are now evaluated from small to large, in the hopes that small values would print faster. (There is no performance test for this claim.)
• The days unit is now printed when total hours is at least 24. (Issue Feature request: formats for long cpu times #1317) There is room in the new code to revert to the old threshold (totalHours < 2400).

[Explorer09]

Examples of all formats:

 0:00.00 - 59:59.99
 1h00:00 - 23h59:59
1d 0h 0m - 9d23h59m
  10d 0h -  364d23h
  1y  0d - 999y364d
   1000y - 9999999y
eternity

[BenBE]

Some suggestions. No perf testing done re mod/div vs. div & mul+sub.

How much of this code could we potentially re-use for #1288 time printing

[Explorer09]

With zero padding the formats will now look like this:

 0:00.00 - 59:59.99
 1h00:00 - 23h59:59
1d00h00m - 9d23h59m
  10d00h -  364d23h
  1y000d - 999y364d
   1000y - 9999999y
eternity - eternity

How much of this code could we potentially re-use for #1288 time printing

Unfortunately I don't think there is potential for code reuse. The time factoring logic would be slightly different between two functions, but at least we can keep variable naming consistent.

[natoscott]

Looks good, though I think that final branch and 'eternity' string is a bit over the top :P (could've just gone up to 9999999y and used that for anything above 1million years).

[Explorer09]

@natoscott Just to say the "eternity" idea was not mine. It was there since commit ec809b7. Credit @BenBE for that one.