[browser] stop testing with nodejs

[pavelsavara]

To reduce WASM test matrix

• removed pipelines
• WasmTestOnNodeJS scenario
• Wasm.Build.Tests tests
• Old Blazor_net50 project that was added in [wasm] Don't use workload for tfm < net6.0 #56606 to be used in WBT but we started using different logic to test Blazor since then and it can be safely removed.

I left in place

• PlatformDetection.IsNotNodeJS
  • HTTP & WS tests
  • Hybrid globalization tests
• detection in `src\mono\browser\test-main.js
• any production code

[ilonatommy]

We might be missing removal of this and places it's being called at:

• runtime/src/mono/wasm/Wasm.Build.Tests/WasmTemplateTestBase.cs

  Line 70 in f944a77

  private static void UpdateRuntimeconfigTemplateForNode(string projectDir)

Edit: it's not called from anywhere...

[pavelsavara]

• UpdateRuntimeconfigTemplateForNode

Is used for wasmconsole template testing, which is V8. Do we want to drop that template too and simplify the host ?

[ilonatommy]

• UpdateRuntimeconfigTemplateForNode

Is used for wasmconsole template testing, which is V8. Do we want to drop that template too and simplify the host ?

We're not testing libs with v8 anymore, so that would make it consistent.
Edit: actually we are, in outerloop and extra-platforms. Sorry, then this argument does not make sense.

[maraf]

Do we want to drop that template too and simplify the host ?

Yes, it can be done here or in a follow up.

Are we going to explicitly remove parts of typescript related to node?

[pavelsavara]

Do we want to drop that template too and simplify the host ?

Yes, it can be done here or in a follow up.

Let's do it in follow up.

@lewing please confirm we have quorum on removing console from dotnet new and dotnet run

Are we going to explicitly remove parts of typescript related to node?

Let's do that in yet another separate PR and we will see how many things there are. I guess it's not that many and possibly we could leave it in place.

[ilonatommy]

I understand that we are using v8 to save resources during running tests, however I don't see much difference in the time used in
/p:Scenario=WasmTestOnChrome and /p:Scenario=WasmTestOnV8, checking biggest lib test project: System.Runtime.Tests (test count differs, so let's look at the averaged value)

V8:     total="64976" time="122.362"
Chrome: total="64970" time="114.317" 
Averaged chome time if it had same number of tests as v8: 122,350

I am proposing:

lane	current scenarios	future scenarios	changed?
LibraryTests Linux	WasmTestOnV8,WasmTestOnChrome,WasmTestOnFirefox	WasmTestOnChrome,WasmTestOnFirefox	Y
LibraryTests Windows	WasmTestOnChrome	WasmTestOnChrome	N
LibraryTests_Smoke_AOT Linux	WasmTestOnV8	WasmTestOnChrome	Y
LibraryTests_Smoke_AOT Windows	WasmTestOnChrome	WasmTestOnChrome	N
LibraryTests_EAT Linux	WasmTestOnV8	WasmTestOnChrome	Y
LibraryTests_Threading Linux	WasmTestOnChrome,WasmTestOnFirefox	WasmTestOnChrome,WasmTestOnFirefox	N
extra-platforms	WasmTestOnV8, WasmTestOnChrome	WasmTestOnChrome	Y
outer-loop	WasmTestOnV8	WasmTestOnChrome	Y

[ilonatommy]

/ba-g failures are DebuggerTests, not related