Try pairing when descriptor write fails

[dakhnod]

The code tries pairing with a BLE device when an initial characteristic read operation fails, and then tries to read the value again after pairing.

Yet, there might be a failing descriptor write operation first, or no characteristic read at all. Hence, this code adds handling of such a case by buffering every write, issuing pairing if needed and re writing the descriptor.

This code is just a starting point. @shmuelzon How would you buffer the writes?

[dakhnod]

@shmuelzon any comment on this?

[dakhnod]

@shmuelzon how about this: Instead of focusing on single operations, we run the regular initialization routine as it used to be. When we hit an insufficient-encryption or encryption we try bonding, and then re-run the regular routine from start. That way, all possible types of read/write operations should be covered.