Errors should respect the standard convention

[BorjaRuizReverter]

I recall that errors should respect the standard convention:

1. Errors of direct measures are equal to the sensibility of the measurement device.
2. Errors of indirect measures should have just 1 or 2 figures, and for having 2 figures errors should be between 10 and 25 (in float notation). From 26 on (in float notation), the error should be rounded to 3.