`Accumulated` computation impl uses constant `computation_id` which may conflict with other ids

`Accumulated` currently implements `Computation` and `Run` itself for users, using a hard-coded computation id:
https://github.com/jfecher/inc-complete/blob/master/src/accumulate.rs#L74
```rs
fn computation_id() -> u32 {
100000
}
```
If the user's computations use this ID - or even if they have more than one accumulated type in their Db this will break.

This should be fixed such that:
1. `Accumulated::::computation_id() != Accumulated::::computation_id()` for `A != B`
2. `Accumulated::::computation_id()` does not conflict with user ids or requires users to assign it an id

Start in src/accumulate.rs at the Computation implementation around line 74, then inspect how computation IDs are assigned and checked for Db users. Define a collision-free approach that gives different Accumulated types distinct IDs without conflicting with user IDs, and verify both stated requirements with tests.

Written by the indexing model from the issue text.