ummm... actually your original calculation that the calendar is wrong by 12 days is actually wrong....
in the calendaring system this is already taken care of the actual forumla for leap years isn't actually
(if year /4 = int) then leapyear = true
in actual fact the formula for working out if it's a leap year is actually
((if year /100 == int)&&(year/400==int)||((year/4=int)&&(year/100!=int)) leapyear = true;
basically if the year is divisible by 100, but is not divisible by 400 it's not a leap year.
if the year is divisible by 4 and not divisible by 100 it's a leap year.
the year 2000 was a leap year (because it's divisible by 100 and 400 without remainder) the yeat 1900 was not a leap year, nor was 1800, 1700, 1500, 1400, 1300, 1100, 900, 700, 600, 500, 300, 200 or 100. (because they are divisible by 100 but not divisible by 400 without remainder)
of course the gregorian calendar was not invented at 0 BC either.
all of the innacuracies that you mentioned are already taken care of....
fear not. march 7th, is still march 7th!
as for the two months to drop... July and August of course
(thus making SEPTember the seventh month, OCTober the 8th, NOVember the 9th and DECember the 10th as originally intended and implied in their names).
anyway... back to the original
a year is 365 days 6 hours 9 minutes 9.54
the real problem here is the fact that the two natural measurements (time taken for earth to orbit the sun, and time taken for earth to revolve once around it's axis) are mutually exclusive events.
with the metric time solution you don't actually fix anything, all you do is make sure that over time midnight appears at midday! and the offset is ignored.
you have to decide what you want to keep, the year measurement of the day measurement. then work out the new si time measurements based on that.
a second is a time measurement based on the time taken for the earth to reveolve around it's axis, (24 * 60 * 60) a year masurement is the time taken for the earth to revolve around the sun, based on
basicall the idea of creating a metric time to tie in the days to standardise the amount of time taken to revolve around the earth is flawed, creating a measurement of base 10 to replace a system doesn't get rid of the fundamental problem that the earth does not have synchronisation between the time taken to revolve and orbit.
you can't tie them together with a standard measurement to get rid of leap years.