Writer node is not returning to readers hostgroup when writerIsAlsoReader is changed

I have 2 nodes, mostly pxc_scheduler_handler default configuration and on proxysql the default split r/w rules:
INSERT INTO mysql_query_rules (rule_id,active,match_digest,destination_hostgroup,apply)
VALUES
(1,1,'^SELECT.*FOR UPDATE',100,1),(2,1,'^SELECT',101,1);

I tested:

root@proxysql1:~# for i in $(seq 1 100); do echo "select @@hostname for update" | mysql -P 6033 -h 127.0.0.1 -u root -ppxc34327 -NA 2>/dev/null; done | sort | uniq -c
    100 node1
root@proxysql1:~# for i in $(seq 1 100); do echo "select @@hostname" | mysql -P 6033 -h 127.0.0.1 -u root -ppxc34327 -NA 2>/dev/null; done | sort | uniq -c
    100 node1

I decided to change:

 writerIsAlsoReader = 1

to:

 writerIsAlsoReader = 0

Nice! it is working:

root@proxysql1:~# for i in $(seq 1 100); do echo "select @@hostname for update" | mysql -P 6033 -h 127.0.0.1 -u root -ppxc34327 -NA 2>/dev/null; done | sort | uniq -c
    100 node1
root@proxysql1:~# for i in $(seq 1 100); do echo "select @@hostname" | mysql -P 6033 -h 127.0.0.1 -u root -ppxc34327 -NA 2>/dev/null; done | sort | uniq -c
    100 node2

But when I reverted (writerIsAlsoReader back to 1), the servers remains as:

+--------------+---------------+------+-----------+--------------+--------+-------------+-----------------+---------------------+---------+----------------+------------------------------------------------------+
| hostgroup_id | hostname      | port | gtid_port | status       | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment                                              |
+--------------+---------------+------+-----------+--------------+--------+-------------+-----------------+---------------------+---------+----------------+------------------------------------------------------+
| 100          | 10.128.15.250 | 3306 | 0         | ONLINE       | 1000   | 0           | 1000            | 0                   | 0       | 0              |     100_W_101_R_retry_up=0;100_W_101_R_retry_down=0; |
| 101          | 10.128.15.253 | 3306 | 0         | ONLINE       | 1      | 0           | 1000            | 0                   | 0       | 0              |                                                      |
| 100          | 10.128.15.253 | 3306 | 0         | OFFLINE_SOFT | 1      | 0           | 1000            | 0                   | 0       | 0              |                                                      |
+--------------+---------------+------+-----------+--------------+--------+-------------+-----------------+---------------------+---------+----------------+------------------------------------------------------+```

The pxc_scheduler_handler conf: [pxc_conf.txt](https://github.com/percona/pxc_scheduler_handler/files/9470182/pxc_conf.txt)
I just want to be sure that it is an expected behavior.

Start with the attached pxc_conf.txt and reproduce the writerIsAlsoReader transition using the shown ProxySQL query rules and runtime_mysql_servers output. Trace how the scheduler handles changing the setting in each direction. Done means establishing whether the retained OFFLINE_SOFT hostgroup entry is expected and documenting or fixing the transition behavior.