Are joins on date data type left outer joins?

I have a custom date dimension table and it is used to calculate a cumulative headcount from my FactHeadcount table. When I join columns of data type “date”, I get the results I expect.  When I join columns on a key “DateKey” of type integer, the Headcount measure breaks.

Other details:

• DimDateFY table has one record for each day between Jan 1, 2015 and Dec 31, 2017
• FactHeadcount table “Effective date” start from 1960 and go to Dec 2016
• I have a report filter selecting the months of Mar 2016 to Feb 2017 from the DimDateFY table
• Headcount =                
      CALCULATE (                                
          SUM (FactHeadcount[Counter]),                                
              FILTER (                                                
                  ALL (DimDateFY[Date]),                                                
                  (DimDateFY[Date]) <= MAX (DimDateFY[Date])                                
              )                
          )
• The only change between behaviors is an update to the relationships

If Power BI joins are inner joins, why is it that it becomes a left outer join if the join is on a data type “date”? Is this expected behavior?  Will it stay this way? 

Is it possible to modify the measure to obtain the correct headcount, with the report filter, and using the “integer” inner join?

If you have advice and/or documentation on this, it would be greatly appreciated.