Fix spurious global xmin > xmin error on rollback

[gfphoenix78]

GetSnapshotData() on a QE computes xmin under ProcArrayLock but reads/advances DistributedLogShared->oldestXmin afterwards, in a different lock domain. In that window a concurrent backend can advance the shared oldestXmin past this snapshot's xmin: when the transaction whose local xid equals xmin aborts, its distributed-log entry keeps distribXid == 0, so DistributedLog_AdvanceOldestXmin()'s forward scan skips over it instead of stopping, bumping oldestXmin beyond xmin. The resulting globalxmin > xmin then tripped a hard elog(ERROR) and failed the query (seen intermittently in the partition_prune regression test).

The aborted xid is invisible to this snapshot anyway, and the visibility horizon only needs to be a valid lower bound, so a horizon at our own xmin loses nothing. Clamp globalxmin to xmin instead of erroring -- always safe, and consistent with the clamp already done in DistributedLog_GetOldestXmin().

Fixes #ISSUE_Number

What does this PR do?

Type of Change

• Bug fix (non-breaking change)
• New feature (non-breaking change)
• Breaking change (fix or feature with breaking changes)
• Documentation update

Breaking Changes

Test Plan

• Unit tests added/updated
• Integration tests added/updated
• Passed make installcheck
• Passed make -C src/test installcheck-cbdb-parallel

Impact

Performance:

User-facing changes:

Dependencies:

Checklist

• Followed contribution guide
• Added/updated documentation
• Reviewed code for security implications
• This PR contains AI-assisted code generation
• Requested review from cloudberry committers

Additional Context

CI Skip Instructions

---

[gfphoenix78]

The problem is:

gpadmin=# create temp table ttmp(a int);
ERROR:  global xmin (1076) is higher than transaction xmin (1075) (procarray.c:3298)  (seg0 127.0.0.1:7002 pid=2466718) (procarray.c:3298)
LINE 1: create temp table ttmp(a int);
                          ^

repro:

-- prepare
drop table if exists t1;

create table t1(a int);

-- T1
1: begin;
1: insert into t1 select i from generate_series(1,10)i;
-- to see the local xid on segments
1: select gp_segment_id, xmin, * from t1;

-- T2, will trigger the error
2: set optimizer=off; -- invoke the QE backend on segments
-- gdb attach to one of the QEs and break after release the ProcArrayLock
-- gdb -p xxxx
-- > b procarray.c:3303
-- > c

-- the following sql will be break when running GetSnapshotData
2: create temp table ttmp(a int);

-- rollback T1
1: rollback;

-- running SQL that advance DistributedLogShared.oldestXmin
1: insert into t1 select i from generate_series(1,10)i;

-- gdb: the QE of T2 should be still hanged in GetSnapshotData
-- quit the gdb now
-- > quit

-- now, the above SQL to create temp table will fail

gdb break the function GetSnapshotData at:

	/*
	 * GP: In computing Globals, also take distributed snapshots into
	 * account.
	 */
	if (TransactionIdPrecedes(xmin, globalxmin))
		globalxmin = xmin;

	if (!IS_QUERY_DISPATCHER())