Adds filamentUsed and longestPrint stats to PrintCounter

[jbrazio]

Fixes #4294.

Stats: Prints: 61, Finished: 46, Failed: 15
Stats: Total time: 2d 13h 6min, Longest job: 0d 8h 23min
Stats: Filament used: 15.19m

[thinkyhead]

Nice. We could also display the total volume of filament. For example, if the filament were an average 2.98mm diameter: π * 1.49^2 * 15.19 gives a total volume of 10.59cc. If there were some way to translate that volume into weight (based on a material profile) it could calculate the material cost of the print too. Someday…

[thinkyhead]

d h min instead of d h m?

Lately I've been favoring d:hh:mm:ss for times, as they are straightforward and compact. Well, actually with M31 I went with…

inline void gcode_M31() {
  millis_t t = print_job_timer.duration();
  int d = int(t / 60 / 60 / 24),
      h = int(t / 60 / 60) % 60,
      m = int(t / 60) % 60,
      s = int(t % 60);
  char time[30];
  if (d)
    sprintf_P(time, PSTR("%i:%02i:%02i:%02i"), d, h, m, s);
  else if (h)
    sprintf_P(time, PSTR("%02i:%02i:%02i"), h, m, s);
  else
    sprintf_P(time, PSTR("%02i:%02i"), m, s);

  lcd_setstatus(time);

  SERIAL_ECHO_START;
  SERIAL_ECHOPGM(MSG_PRINT_TIME " ");
  SERIAL_ECHOLN(time);

  thermalManager.autotempShutdown();
}

I guess there could be some confusion with mm:ss because it could be interpreted as hh:mm. So maybe the d h m s convention is more clear. I do think it's fine to leave off 0d, 0h, etc., for any of the time components that are zero. So you could see 1d 25s.

[thinkyhead]

I've gone back to the 1d 2h 3m 4s format. And in fact, I'm thinking we can replace the time report at the end of SD printing with a simple call to M31.

[jbrazio]

How about a datetime struct that we could reuse everywhere ?

  datetime_t t = print_job_timer.duration();

  /*
  Example usage:
  int d = t.days();
      h = t.hours(),
      m = t.minutes(),
      s = t.seconds();
  */

  lcd_setstatus(t.toString());
  SERIAL_ECHO_START;
  SERIAL_ECHOPGM(MSG_PRINT_TIME " ");
  SERIAL_ECHOLN(t.toString());

[thinkyhead]

So long as it's reasonably economical compared to the existing code.

[RDickens]

For the filament used counter I was actually thinking of a running total so you could see how many hundreds of meters has been used. This would be handy to help plan service intervals on my printer and reduce down time.

[jbrazio]

"Filament used" is an accumulative counter.

[thinkyhead]

Is millis() a volatile function? If so, this will compile 2 function calls. Otherwise it will init in one go.