Unexpected data in Blob Dataset 2020

The schema of the 2020 Blob Dataset presents `AnonFunctionInvocationId` and `AnonAppName` as unique IDs.

However, there are sometimes discrepancies where the invocationId spans multiple application names. For example,

```
full_df[full_df['AnonFunctionInvocationId'] == 1967128581]
```

Timestamp | AnonRegion | AnonUserId | AnonAppName | AnonFunctionInvocationId | AnonBlobName | BlobType | AnonBlobETag | BlobBytes | Read | Write | Datetime
-- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | --
1606814873193 | q2d | 1209884869 | 01qqaww4 | 1967128581 | 1wx5dgohq1kiwjum | BlockBlob/text/plain; charset=utf-8 | f1x5p2nqh6 | 28.0 | True | False | 2020-12-01 09:27:53.193
1607004493391 | q2d | 1209884869 | j2alqt8s | 1967128581 | 1wx5dgohq1kiwjum | BlockBlob/text/plain; charset=utf-8 | w5mohi6523 | 28.0 | True | False | 2020-12-03 14:08:13.391

This seems to be a recurrent pattern with this user, for example consider other `functionInvocationIds` 830734703, 440926898, or 900464655.

This leads me to believe that the cause is unlikely to be unfortunate prefixes of hashed IDs. Is there any way to explain this discrepancy, apart from the data being potentially unclean?