I am able to replicate this observation. The cases with MIGRATE1==1 and MOVEDIN==0 can be explained by the fact that MOVEDIN is only asked of householders (RELATE==1). All other individuals within the household are given a MOVEDIN value of 0, regardless of their value recorded in MIGRATE1. The remaining cases with MIGRATE==1 and MOVEDIN==1 represent less than 1% of the samples. Therefore, it seems likely these observations are driven by respondent error or due to question ambiguity.
Note that the question for MIGRATE1 asks respondents: “Did this person live in this house or apartment 1 year ago?” The question for MOVEDIN asks respondents: When did PERSON 1 (listed on page 2) move into this house, apartment, or mobile home?" Therefore, it is seemingly possible for someone to have moved into their house 1 year ago, and answer that they lived in the same house 1 year ago (via MIGRATE1) and that they moved 12 months ago or less (via MOVEDIN).
Finally, note the data quality flags for MOVEDIN, QMOVEDIN, which may further clarify these cases.