Exercise 2: Trust Policy vs. Permission Policy — Possible Solution
====================================================================

The difference, per the chapter:

TRUST POLICY: attached to the TARGET role (the role being assumed),
this controls WHO is allowed to assume it in the first place -- which
accounts, users, or services are permitted to even begin the
role-assumption process.

PERMISSION POLICY: also attached to the role, this controls WHAT the
assumer can actually DO once they've successfully assumed it -- the
actual set of allowed/denied actions and resources available to
someone operating as that role.

What a trust policy misconfiguration looks like as a symptom:

The role-ASSUMPTION step itself fails -- the requester is told
directly that they are not permitted to assume the role at all. This
happens BEFORE any question of what actions they'd be allowed to
perform even arises, because they never successfully become that
role in the first place. The error occurs at the very first step of
the process.

What a permission policy misconfiguration looks like as a symptom:

Role assumption SUCCEEDS -- the requester successfully becomes the
role -- but subsequent actions attempted while operating as that role
are denied. This looks like an ordinary "access denied" error on a
SPECIFIC action, occurring only AFTER the assumption step already
completed without any problem, which is the key symptom distinguishing
it from a trust policy failure.

Why the distinction matters practically: if someone troubleshooting
this sees an assumption-step failure, checking or modifying the
PERMISSION policy is the wrong place to look entirely -- the problem
never gets that far. Conversely, if assumption succeeds but a specific
action is denied, the trust policy was never the issue, and adjusting
it wouldn't fix anything.

WHY THIS WORKS AS AN ANSWER
------------------------------
This distinguishes the two policies by WHERE IN THE PROCESS each
failure would actually manifest -- at the assumption step itself
(trust policy) versus after a successful assumption, on a specific
subsequent action (permission policy) -- which is exactly the
practical, symptom-based distinction the chapter's own warn-box is
built around, and directly explains why confusing the two "wastes real
troubleshooting time."