DateDifference inconsistency - feature or bug?

The following code attempts to measure the difference in months between the last day of a year and the first day of a month in the following year.

QuantityMagnitude@DateDifference[{2018, 12, 31}, {2019, #, 1}, "Month"] & /@ Range@12

which produces

{0.0322581, 1.03571, 2.03571, 3.03333, 4.03333, .03333,
6.03333, 7.03226, 8.03333, 9.03333, 10.0333, 11.0333}

However, the results are not what I would expect. Specifically, I would expect the denominator in calculating the fractional part to be the number of days in that particular month, i.e.

1/{31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31} // N

{0.0322581, 0.0357143, 0.0322581, 0.0333333, 0.0322581, 0.0333333, 0.0322581, 0.0322581, 0.0333333, 0.0322581, 0.0333333, 0.0322581}

What appears to be used instead is a sequence of days in each month of the year which is;

31, 28, 28, 30, 30, 30, 30, 31, 30, 30, 30, 30

which doesn’t make much sense. Not only that, but it leads to nonsensical results such as the following;

In[359]:= DateDifference[{2018, 12, 31}, {2019, 3, 28}, "Month"]

Out[359]= Quantity[3, "Months"]

In[360]:= DateDifference[{2018, 12, 31}, {2019, 3, 29}, "Month"]

Out[360]= Quantity[3.03571, "Months"]

In[361]:= DateDifference[{2018, 12, 31}, {2019, 3, 30}, "Month"]

Out[361]= Quantity[3.07143, "Months"]

In[362]:= DateDifference[{2018, 12, 31}, {2019, 3, 31}, "Month"]

Out[362]= Quantity[3, "Months"]

Is this a bug or am I missing something? I can’t see how this would be a feature.

Many thanks for your thoughts.

Denominator@Rationalize@
    DateDifference[{2018, 12, 31}, {2019, #, 1}, "Month"][[1]] & /@ Range@12

dateDifference[from_, to_, u:"Month"] :=
 DateDifference[from + {0, 0, 1}, to + {0, 0, 1}, u]

Denominator@Rationalize@
    dateDifference[{2018, 12, 31}, {2019, #, 1}, "Month"][[1]] & /@ Range@12