fix(go/adbc/pkg): follow CGO rules properly

[lidavidm]

Depends on apache/arrow#36670.

Fixes #729.

[lidavidm]

TODOs:

• Update to upstream PR
• Move the new test to be specific to the Go driver

[lidavidm]

There's an apparent leak from the new test, will have to look

[zeroshade]

this all LGTM but i don't see any obvious issues that cause the leaked memory

[lidavidm]

Weirdly enough, it's in an old test:

Note: Google Test filter = SqliteFlightSqlConnectionTest.MetadataGetTableTypes
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from SqliteFlightSqlConnectionTest
[ RUN      ] SqliteFlightSqlConnectionTest.MetadataGetTableTypes
[       OK ] SqliteFlightSqlConnectionTest.MetadataGetTableTypes (5 ms)
[----------] 1 test from SqliteFlightSqlConnectionTest (5 ms total)

[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (6 ms total)
[  PASSED  ] 1 test.

=================================================================
==1566443==ERROR: LeakSanitizer: detected memory leaks

Direct leak of 128 byte(s) in 2 object(s) allocated from:
    #0 0x7f4c6193017a in __interceptor_calloc ../../../../libsanitizer/asan/asan_malloc_linux.cpp:154
    #1 0x7f4c611c0597 in _cgo_0a637ba8d569_C2func_calloc (/home/lidavidm/Code/upstream/adbc/adbc-crash/build/driver/flightsql/libadbc_driver_flightsql.so.6+0x988597)
    #2 0x7f4c60b34fc0 in runtime.asmcgocall.abi0 (/home/lidavidm/Code/upstream/adbc/adbc-crash/build/driver/flightsql/libadbc_driver_flightsql.so.6+0x2fcfc0)

SUMMARY: AddressSanitizer: 128 byte(s) leaked in 2 allocation(s).
fish: Job 1, './driver/flightsql/adbc-driver-…' terminated by signal SIGABRT (Abort)

[lidavidm]

Ugh...moving the arrow-go version back fixes it. What happened...

[lidavidm]

ah, I think I got it - PR incoming?

[lidavidm]

Alright, I'm too frazzled to figure out the right thing today but since the other PR added a Retain() call on exported readers, we need to remember to Release() here now

[lidavidm]

Actually no I just copy-pated a line twice, that's why my fix wasn't working

That should hopefully do it...

[lidavidm]

And apache/arrow#36676 just for sanity

[lidavidm]

hmm, my local clang-format seems to disagree with github all of a sudden...

[zeroshade]

LGTM awesome