Has anyone had problems with Sentinel 2 data not overlying exactly when looking at different image dates? Some scenes appear to be moving by at least one pixel (e.g. in Bands 2, 3 and 4). I am using L1C data.
In the first few months (6 or so) of Sentinel-2B it was a known issue that there is a shift, along the lines of what you are describing.
This blog post describes some of the problems other users have had with multi-temporal registration for Sentinel 2 data: http://www.cesbio.ups-tlse.fr/multitemp/?p=13112
From the same author, there are reports of plans to improve the registration of archival data: