Proposal for Changing index request project_id filter

From the past to today, it seems that Octavia has a different project_id filter process than other Openstack services (Nova, Cinder, ETC).
According to the documentation at https://docs.openstack.org/octavia/latest/configuration/policy.html, I believe that the get_read-global action in octavia is the same function as the get_all_tenants action in Nova and has the same RBAC rule.
However, the implementation of the List API is different from Nova.

First, Octavia is calling the filter calculation method from a single list api entry point.
(https://github.com/openstack/octavia/blob/master/octavia/api/v2/controllers/load_balancer.py#L86)
Next, if the RBAC_GET_ALL_GLOBAL Rule is allowed, no project_id filter is added for the current context.
(https://github.com/openstack/octavia/blob/master/octavia/api/v2/controllers/base.py#L226-L228)
Alternatively, RBAC_GET_ALL_GLOBAL rules must be disallowed, but check for the presence of RBAC_GET_ALL rules by referencing the project_id of the context.
(https://github.com/openstack/octavia/blob/master/octavia/api/v2/controllers/base.py#L230-L240)

I consider the all_global action of admin to be optionally supported, and I think there should be a functional delimiter of “all_tenants” similar to Nova, what do you think?
I'm not comfortable with contexts with admin roles causing mistakes or only looking at the context's project loadbalancer.

I think it would be fun to do a feature together if you don't mind.