Debug windows e2e tests

[rly]

I thought we had fixed it. I think the core issue is on Windows, generating the test data takes longer than ~3 minutes and the process times out.

See https://github.com/NeurodataWithoutBorders/nwb-guide/actions/runs/10527059741/job/29169359993#step:13:484
and https://github.com/NeurodataWithoutBorders/nwb-guide/actions/runs/10527059741/job/29169359993#step:13:611

The last PCA output shows:

stdout | tests/e2e/tutorial.test.ts > E2E Test > Run through several pipeline workflows > Complete a multi-session workflow > Specify data formats
[electron] 
Fitting PCA:  58%|#####8    | 29/50 [02:54<02:52,  8.22s/it]

After that it says:

stdout | tests/e2e/tutorial.test.ts
[electron] [2024-08-23 14:12:59,405] INFO in app: Shutting down server

and

 FAIL  tests/e2e/tutorial.test.ts > E2E Test > Create tutorial dataset
ProtocolError: Runtime.callFunctionOn timed out. Increase the 'protocolTimeout' setting in launch/connect calls for a higher timeout if needed.
 ❯ Callback.<instance_members_initializer> node_modules/puppeteer-core/src/common/CallbackRegistry.ts:114:12
 ❯ new Callback node_modules/puppeteer-core/src/common/CallbackRegistry.ts:119:3
 ❯ CallbackRegistry.create node_modules/puppeteer-core/src/common/CallbackRegistry.ts:27:22
 ❯ Connection._rawSend node_modules/puppeteer-core/src/cdp/Connection.ts:125:22
 ❯ CdpCDPSession.send node_modules/puppeteer-core/src/cdp/CDPSession.ts:95:29
 ❯ ExecutionContext.#evaluate node_modules/puppeteer-core/src/cdp/ExecutionContext.ts:417:44
 ❯ ExecutionContext.evaluate node_modules/puppeteer-core/src/cdp/ExecutionContext.ts:292:32
 ❯ IsolatedWorld.evaluate node_modules/puppeteer-core/src/cdp/IsolatedWorld.ts:194:26
 ❯ CdpFrame.evaluate node_modules/puppeteer-core/src/api/Frame.ts:484:35
 ❯ CdpFrame.<anonymous> node_modules/puppeteer-core/src/util/decorators.ts:63:21

[rly]

The Windows e2e tests run now. It is still not clear to me why the "Fitting PCA" step of generating the test data takes ~ 6 minutes on Windows CI but only a few seconds on Linux and Mac. I wonder if that is the case on typical Windows machines in the wild. @CodyCBakerPhD do you have a Windows machine and can verify that?

[CodyCBakerPhD]

@CodyCBakerPhD do you have a Windows machine and can verify that?

Now that I'm back home, I do - but on my machines it works very quickly. Probably just something inefficient in the GitHub architecture side for that platform

[CodyCBakerPhD]

As always, thanks for tackling this!