The existing schema_abort csuite test covers schema operations under crash and recovery for ASC. This ticket adds equivalent coverage for a disaggregated leader.

Write a new csuite test modeled on schema_abort, adapted for disagg:

Workload:

• Open a WT_CONNECTION with disaggregated storage (type=layered tables).
• Run N worker threads that create and drop layered tables from a fixed URI pool, each operation assigned a value from a shared monotonically increasing schema_op_epoch counter. Each thread appends "CREATE/DROP <epoch> <uri>" to a per-thread record file on disk.
• Run a checkpoint thread that advances stable_disaggregated_schema_epoch to the current schema_op_epoch before each checkpoint, triggering WiredTiger to include all schema operations up to that epoch in the checkpoint.
• Parent process sends SIGKILL to the child after at least one checkpoint completes.

Verification:

• After recovery, query last_disaggregated_schema_epoch from the connection. This is the epoch cutoff: all schema operations at or below this value are durable; operations above it may not be.
• Replay the per-thread record files up to the cutoff. For each URI slot, find the most recent operation with epoch <= cutoff. If it was a CREATE, assert the table exists and is readable. If it was a DROP, assert the table is absent.
• For each surviving table, verify that the data row written at the latest epoch <= cutoff is present with the correct value.

Deliverables:

• New csuite test under test/csuite/schema_disagg_leader_abort/ with at minimum: main.c (parent/orchestration), workload.c (worker and checkpoint threads), verify.c (post-recovery verification).
• CMakeLists.txt entry wiring the new binary into the csuite build.
• Evergreen task entry in test/evergreen_disagg.yml running the test on the disagg build variant, including a stress variant with extended run time.