Replace boolean mask in particale structure with integer

[Sichao25]

Fix part of issue #156 and #119 . As reported, the boolean mask implementation became corrupted when used with a large number of particles. Switching to an integer mask resolves this issue.

[Sichao25]

@zhangchonglin Replacing the Boolean mask with an integer resolves the code crash with Cabana-based data structures in my pumipic tests. Would you mind testing this change in your code to verify it helps when you have time? As a temporary fix, you could simply change AppendMT<bool,DataTypes> here to AppendMT<int,DataTypes> instead of merging this entire branch.

[jacobmerson]

@zhangchonglin did this PR resolve your issue?

[zhangchonglin]

@jacobmerson: this solved the particle information corruption/code crash issue for Comet. I assume this will solve similar issue in XGCm, etc.

@Sichao25: were you able to figure out what is the reason behind the particle information corruption/crash. I saw Sam Reeve's comment for your issue ECP-copa/Cabana#842. Is this related to ECP-copa/Cabana#579, due to AoSoA type ordering?

[Sichao25]

@zhangchonglin It could be similar padding issue, but probably not the same one because we've already using the correct order.