Do not cancel update index store tasks in favor of a task with fewer files

We should not take the number of files in an `UpdateIndexStoreTaskDescription` as an indication on how important the task is. If we do need this functionality, eg. because we want to update the index of files with syntactic matches for a rename term, this should be communicated using a specific purpose similar to `TargetPreparationPurpose`. Since the only reason we update the index store for a file right now is background indexing, such a check is not needed.

Author: ahoppen

Sources/SKTestSupport/SkipUnless.swift

@@ -476,7 +476,7 @@ package actor SkipUnless {
     file: StaticString = #filePath,
     line: UInt = #line
   ) async throws {
-    return try await shared.skipUnlessSupportedByToolchain(swiftVersion: SwiftVersion(6, 2), file: file, line: line) {
+    return try await shared.skipUnlessSupportedByToolchain(swiftVersion: SwiftVersion(6, 3), file: file, line: line) {
       return await ToolchainRegistry.forTesting.default?.canIndexMultipleSwiftFilesInSingleInvocation ?? false
     }
   }

Sources/SemanticIndex/UpdateIndexStoreTaskDescription.swift

@@ -242,15 +242,7 @@ package struct UpdateIndexStoreTaskDescription: IndexTaskDescription {
         // Disjoint sets of files can be indexed concurrently.
         return nil
       }
-      if self.filesToIndex.count < other.filesToIndex.count {
-        // If there is an index operation with more files already running, suspend it.
-        // The most common use case for this is if we schedule an entire target to be indexed in the background and then
-        // need a single file indexed for use interaction. We should suspend the target-wide indexing and just index
-        // the current file to get index data for it ASAP.
-        return .cancelAndRescheduleDependency(other)
-      } else {
-        return .waitAndElevatePriorityOfDependency(other)
-      }
+      return .waitAndElevatePriorityOfDependency(other)
     }
   }