Add a common prefix to all pipeline stages that aren't injected by 1ES pipeline templates

[lbussell]

When reviewing a pipeline's output, it's difficult to determine which stages we control and which we don't. Lots of additional stages are injected by required pipeline templates.

docker-tools/eng/common/templates/1es-official.yml

Lines 40 to 60 in f4c1d80

	extends:
	template: v1/1ES.Official.PipelineTemplate.yml@1ESPipelineTemplates
	parameters:
	pool: ${{ parameters.pool }}
	sdl:
	binskim:
	enabled: true
	componentgovernance:
	ignoreDirectories: $(Build.SourcesDirectory)/versions
	whatIf: ${{ parameters.cgDryRun }}
	showAlertLink: true
	policheck:
	enabled: true
	sourceRepositoriesToScan:
	exclude:
	- repository: InternalVersionsRepo
	- repository: PublicVersionsRepo
	sourceAnalysisPool: ${{ parameters.sourceAnalysisPool }}
	tsa:
	enabled: true
	stages: ${{ parameters.stages }}

We should add some logic to inject a common prefix to each task/step's displayName. Something simple that stands out is an emoji like 🟪. (purple because that's .NET's color).

One problem that I see with this approach is that the postPublishNotification ImageBuilder command relies on the displayName of tasks to check if they were successful.

docker-tools/eng/common/templates/jobs/publish.yml

Lines 199 to 208 in f4c1d80

	--task "Copy Images (Authenticated)"
	--task "Publish Manifest (Authenticated)"
	--task "Wait for Image Ingestion (Authenticated)"
	--task "Publish Readmes"
	--task "Wait for MCR Doc Ingestion (Authenticated)"
	--task "Publish Image Info"
	--task "Ingest Kusto Image Info (Authenticated)"
	--task "Generate EOL Annotation Data (Authenticated)"
	--task "Annotate EOL Images (Authenticated)"
	--task "Wait for Annotation Ingestion (Authenticated)"

That behavior should be based on the name of the steps, not the displayName. That way we can do whatever we want with displayNames without worrying about breaking publish notifications.