Get positions script should not panic

[illia-malachyn]

What is done

getPositionDetails panics when called with a closed/non-existent position ID.
This is problematic for batch queries (get_positions_by_ids.cdc) - one stale ID
causes the entire batch to fail.

Refactor _borrowPosition to delegate to a new _tryBorrowPosition that returns
an optional reference instead of panicking. Add tryGetPositionDetails that
guards on existence before delegating to getPositionDetails.

Update get_positions_by_ids.cdc to use tryGetPositionDetails so closed positions
are silently skipped instead of failing the batch.

Why

The fcm-observer queries positions in batches using get_positions_by_ids.cdc. Here's the problem:

1. We call get_position_ids.cdc -> get back [0, 3, 7, 12, ...]
2. We call get_positions_by_ids.cdc with those IDs
3. Between step 1 and step 2, position 7 gets closed on-chain
4. getPositionDetails(pid: 7) panics -> the entire batch fails, all 50 positions get no update

We're always behind - by the time we use the IDs, some may already be stale. tryGetPositionDetails skips missing positions instead of panicking, so the rest of the batch succeeds.