[ARM] Make arm encoder init more robust

Related Issues and PRs

If we happen to robot init with arm in slop zone we got bad motor init. Also if the encoder wasn't working at all we got bad results.

Issue:

Related PRs:

[x] Code is complete, and naming of variables, methods and classes are readable by non-coder humans.
[x] This code change has been tested and validated working as expected.
[ ] This code change has been reflected in robot documentation in the docs folder.
[x] The questions below have all been answered.
[x] The title of this PR has been updated to describe the change made, for example, [Subsystem or Command Name] Short description of change.
[x] Assignees (usually yourself) have been assigned in the "Assignees" section to the right.
[x] Relevant reviewers have been assigned under the "Reviewers" section at the right.

Defined a top and bottom of each slop zone
Arm.java separately accounts for each zone including slop zones and record a more complete status
Arm.java assumes in midpoint of slop if in a slop zone at init
If already passed an init and motor hasn't reset we keep old motor encoder setting as its probably better than what we get now
If cancoder doesn't read at all at init we assume Auto hardware setup arm position

Tested on robot.