ChipLog* no longer logs to stdout/stderr in darwin-framework-tool

[bzbarsky-apple]

This is fallout from PR #23528.

Before that change, running something like darwin-framework-tool tests TestCluster will show these log bits:

[1671725279532] [76381:6799754] CHIP: [TOO]  **** Test Start: TestCluster
[1671725279532] [76381:6799754] CHIP: [TOO]  ***** Test Step 0 : Wait for the commissioned device to be retrieved
[1671725279532] [76381:6799755] CHIP: [TOO]  ***** Test Step 1 : Send Test Command

before it stops progress due to trying to find the server to send the server to send the command to and failing. After that change, those messages are not logged.

The problem, as far as I can tell, is that darwin-framework-tool links two copies of CHIPLogging.cpp: one as part of Matter.framework and one directly in the executable.

Which means there are two copies of sLogRedirectCallback. darwin-framework-tool sets up (indirectly, via calling MTRSetLogCallback on the framework) a log redirect that logs to stdout.

Now if we have a ChipLog* call from inside the SDK/framework it ends up in chip::Logging::Logv, it will find the framework copy of sLogRedirectCallback, that's not null, it will get called, and the output will appear on stdout.

But if we hav a ChipLog* call from directly inside darwin-framework-tool (e.g. the generated test code that logs the things above), then it will find the other copy of sLogRedirectCallback, which is null. It will then call Platform::LogV, which is a no-op on Darwin, and the log gets lost.