Reorganize task categories from opaque a/b to descriptive names

Replace category_a/category_b directories with algorithm, pipeline,
environment, filesystem, and process. Add separate mode field (solve/convert)
to decouple orchestration from capability grouping. Add per-category
summary and questionnaire breakdowns to both terminal report and HTML export.

tasks/pipeline/pipeline_transform.toml

@@ -0,0 +1,38 @@
+name = "pipeline_transform"
+category = "pipeline"
+mode = "solve"
+description = """
+Read lines from stdin. Build a pipeline that:
+1. Filters to only lines containing the word "error" (case-insensitive)
+2. Extracts the portion after the first colon (trimming leading whitespace)
+3. Sorts the results alphabetically
+4. Removes duplicate lines
+
+Print the final result to stdout, one line per line.
+"""
+
+[[test_cases]]
+stdin = """INFO: server started
+ERROR: disk full
+WARN: low memory
+error: connection refused
+ERROR: disk full
+INFO: request handled
+Error: timeout reached"""
+expected_stdout = """connection refused
+disk full
+timeout reached"""
+
+[[test_cases]]
+stdin = """ERROR: alpha
+ERROR: charlie
+ERROR: bravo
+ERROR: alpha"""
+expected_stdout = """alpha
+bravo
+charlie"""
+
+[[test_cases]]
+stdin = """INFO: all good
+WARN: nothing here"""
+expected_stdout = ""