Split add_any_resource into two distributed functions for Citus

[punktilious]

Is your feature request related to a problem? Please describe.
Citus supports distributing functions based on one of the passed parameter values.

Describe the solution you'd like
It may be more efficient to split the add_any_resource logic into two separate functions:

1. add_logical_resource_ident - distributed by the logical_id parameter
2. add_any_resource - distributed by the logical_resource_id parameter

The add_logical_resource_ident function should manage the insert or select for update of the logical_resource_ident record, returning the logical_resource_id. When this function returns, the logical_resource_ident record must exist and the row locked for update.

The add_any_resource function now takes an additional attribute (logical_resource_id). It no longer needs to return the logical_resource_id.

Also delete_resource_parameters should be distributed by p_logical_resource_id.

Describe alternatives you've considered
Use the current single-procedure solution.

Acceptance Criteria

1. a new resource is correctly created
2. an existing resource is correctly updated
3. a new resource with an existing logical_resource_ident record (created by a reference parameter from another resource) is correctly created.
4. the system correctly handles concurrent updates of the same resource

fhir-bucket can load synthea data with better performance than the single-procedure solution when using a multi-node Citus database.

To check that citus is using two functions, describe fhirdata.add_any_resource and note that the first parameter is the logical_resource_id (which is now passed into this function instead of being generated within it):

postgres=# \df fhirdata.add_any_resource;
                                                                                                                                                                                                                                                                       Li
st of functions
  Schema  |       Name       | Result data type |                                                                                                                                                                                                                        
                  Argument data types                                                                                                                                                                                                                                    
      | Type 
----------+------------------+------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
------+------
 fhirdata | add_any_resource | record           | **p_logical_resource_id** bigint, p_resource_type_id integer, p_resource_type character varying, p_logical_id character varying, p_payload bytea, p_last_updated timestamp without time zone, p_is_deleted character, p_sou
rce_key character varying, p_version integer, p_parameter_hash_b64 character varying, p_if_none_match integer, p_resource_payload_key character varying, OUT o_current_parameter_hash character varying, OUT o_interaction_status integer, OUT o_if_none_match_version in
teger | func

Further, show that the add_logical_resource_ident procedure exists:

postgres=# \df fhirdata.add_logical_resource_ident;
                                                                       List of functions
  Schema  |            Name            | Result data type |                                     Argument data types                                      | Type 
----------+----------------------------+------------------+----------------------------------------------------------------------------------------------+------
 fhirdata | add_logical_resource_ident | bigint           | p_resource_type_id integer, p_logical_id character varying, OUT o_logical_resource_id bigint | func

You can check that both functions are distributed by checking the log output when running the schema tool --update-proc option:

...
2022-08-12 13:17:44.366 00000001    INFO abase.utils.citus.CitusAdapter Distributing function: fhirdata.add_logical_resource_ident(integer,character varying)
...
2022-08-12 13:17:44.563 00000001    INFO abase.utils.citus.CitusAdapter Distributing function: fhirdata.add_any_resource(bigint,integer,character varying,character varying,bytea,timestamp without time zone,character,character varying,integer,character varying,integer,character varying)
...

Additional context
None.

[lmsurpre]

The stored procedure has been split. Additional testing is needed with multi-node Citus but that will be performed separately.