[CLIENT-1337] Unmask C client errors

[juliannguyen4]

Build wheels all passes: https://github.com/aerospike/aerospike-client-python/actions/runs/6240538285

Valgrind
Stage branch: https://github.com/aerospike/aerospike-client-python/actions/runs/6227886001
This branch: https://github.com/aerospike/aerospike-client-python/actions/runs/6240538715

jnguyen@as-PF3WN8Y6:~/aerospike/aerospike-client-python$ diff <(grep Aerospike stage.log | awk '{print $5}' | sort) <(grep Aerospike 'CLIENT-1337.log' | awk '{print $5}' | sort)
5a6,9
> AerospikeClient_Admin_Create_Role
> AerospikeClient_Admin_Create_Role
> AerospikeClient_Admin_Create_Role
> AerospikeClient_Admin_Create_User
13a18
> AerospikeClient_Admin_Drop_User
74a80
> AerospikeClient_New
105a112
> AerospikeClient_Remove
108a116,117
> AerospikeClient_Remove_Invoke
> AerospikeClient_Select
112a122
> AerospikeClient_Select_Invoke
124a135
> AerospikeClient_Type_Init
218d228
< AerospikeQuery_Get_Partitions_status

Extra leaks in this branch

All extra leaks in this branch come from raise_exception or raise_exception_old. This is because the errors from the C client are no longer being reset by the Python client, so I believe more as_error fields are being converted to Python objects and stored in the relevant exception class. Since the class isn't cleaned up properly, this causes more memory leaks from the class's attributes.

3 AerospikeClient_Admin_Create_Role leaks from raise_exception
1 AerospikeClient_Admin_Create_User leak from raise_exception

AerospikeClient_New:

Stage

5 from raise_exception
4 from AerospikeClient_Type_Init

This branch

5 from raise_exception
5 from AerospikeClient_Type_Init

1 AerospikeClient_Remove leak from raise_exception_old
1 AerospikeClient_Select leak from raise_exception_old

AerospikeClient_Type_Init:

Stage

5 from raise_exception
4 from as_config_add_host

This branch

5 from as_config_add_host
5 from raise_exception

[codecov-commenter]

Codecov Report

Patch coverage: 100.00% and no project coverage change.

Comparison is base (39d0759) 81.30% compared to head (1a9338d) 81.31%.

❗ Current head 1a9338d differs from pull request most recent head ee96cc2. Consider uploading reports for the commit ee96cc2 to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##            stage     #470   +/-   ##
=======================================
  Coverage   81.30%   81.31%           
=======================================
  Files          98       98           
  Lines       14808    14757   -51     
=======================================
- Hits        12039    11999   -40     
+ Misses       2769     2758   -11     

Files Changed	Coverage Δ
src/main/client/admin.c	81.29% <ø> (-0.12%)	⬇️
src/main/client/apply.c	86.02% <ø> (-0.15%)	⬇️
src/main/client/exists.c	92.85% <ø> (-0.11%)	⬇️
src/main/client/get.c	95.38% <ø> (-0.07%)	⬇️
src/main/client/operate.c	91.45% <ø> (-0.03%)	⬇️
src/main/client/put.c	97.18% <ø> (-0.08%)	⬇️
src/main/client/query.c	63.24% <ø> (+0.08%)	⬆️
src/main/client/remove.c	83.82% <ø> (-0.47%)	⬇️
src/main/client/remove_bin.c	84.40% <ø> (-0.42%)	⬇️
src/main/client/select.c	90.09% <ø> (-0.10%)	⬇️
... and 2 more

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dwelch-spike]

I haven't taken a very close look but my guess about the new memory leaks @juliannguyen4 mentioned is that they are coming from the new calls to raise_exception() that are hit because err.code isn't reset in many places now.

[dwelch-spike]

LGTM