Duplicate IDs when using reject-sampling

[csala]

Environment Details

Please indicate the following details about the environment in which you found the bug:

• SDV version: 0.8.0
• Python version: any
• Operating System: any

Error Description

When a tabular sampling call involves reject-sampling, the ids are reset at each re-sampling call, potentially producing duplicate ids.

Steps to reproduce

In [1]: import sdv

In [2]: users = sdv.demo.load_demo()['users']

In [3]: model = sdv.tabular.CTGAN(primary_key='user_id', epochs=10)

In [4]: model.fit(users)

In [5]: model.sample(10, conditions={'gender': 'M'})
Out[5]: 
   user_id country gender  age
0        0      FR      M   19
1        4      FR      M   69
2        5      US      M   66
3        8      ES      M   62
4        9      DE      M   54
5        2      BG      M   44
6        3      ES      M   64
7        5      DE      M   61
8        1      US      M   52
9        1      US      M   65

Steps to fix

To fix this, any id fields should be re-populated after all the rows have been successfully sampled.