CBG-4345 don't panic if role documents can't be read

[torcolvin]

CBG-4345 don't panic if role documents can't be read

This is behind #7814

Pre-review checklist

• Removed debug logging (fmt.Print, log.Print, ...)
• Logging sensitive data? Make sure it's tagged (e.g. base.UD(docID), base.MD(dbName))
• Updated relevant information in the API specifications (such as endpoint descriptions, schemas, ...) in docs/api

Dependencies (if applicable)

• Link upstream PRs

Integration Tests

• GSI=true,xattrs=true https://jenkins.sgwdev.com/job/SyncGatewayIntegration/136/

[Copilot]

Pull Request Overview

This PR addresses panic issues when role documents cannot be read by converting functions that previously panicked into error-returning functions. The changes ensure graceful error handling throughout the authentication and channel access system, particularly when role documents are missing or corrupted.

Key changes:

• Modified authentication functions to return errors instead of panicking when role documents can't be read
• Updated all callers to handle the new error returns
• Added proper error logging and user-friendly error responses

Reviewed Changes

Copilot reviewed 22 out of 22 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
auth/principal.go	Updated interface definitions to return errors for channel access methods
auth/user.go	Modified user methods to return errors when role loading fails instead of panicking
auth/role.go	Updated role methods to return errors for channel access operations
auth/user_collection_access.go	Added error handling to collection-specific channel access methods
auth/role_collection_access.go	Updated collection role methods to return errors
auth/collection_access.go	Modified interface definitions for collection access to include error returns
db/crud.go	Updated MakeUserCtx to handle errors from user channel operations
db/changes.go	Added error handling throughout changes feed processing
db/design_doc.go	Updated view filtering to handle channel access errors
db/functions/function.go	Added error handling for channel authorization in functions
db/functions/js_function.go	Updated JavaScript function execution to handle user context errors
rest/handler.go	Modified audit logging to handle role retrieval errors gracefully
rest/bulk_api.go	Updated bulk operations to handle channel access errors properly
rest/admin_api.go	Added error handling to principal marshaling
rest/diagnostic_api.go	Updated diagnostic API to handle channel access errors
rest/user_api_test.go	Updated tests to handle new error returns
rest/role_api_test.go	Modified tests to check for errors in channel operations
db/util_testing.go	Added error handling to test utilities
db/database_test.go	Updated tests to handle new error signatures
auth/user_test.go	Modified tests to check for errors in user operations
auth/auth_test.go	Updated authentication tests to handle new error returns
auth/collection_access_test.go	Added error checking to collection access tests

[Copilot]

The panic replacement uses base.AssertfCtx followed by returning an error. This is inconsistent - either use the assertion (which may panic in debug builds) or return the error, but not both. Consider removing the assertion and only returning the HTTP error.

Suggested change

base.AssertfCtx(h.ctx(), "User %q has no channels in handleAllDocs", base.UD(h.user.Name()))