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/filesystem/multi_file_search.toml

@@ -0,0 +1,30 @@
+name = "multi_file_search"
+category = "filesystem"
+mode = "solve"
+description = """
+You are given a working directory containing several text files.
+Read a search pattern (a simple string, not regex) from stdin.
+Search all .txt files in the working directory for lines containing that pattern (case-sensitive).
+
+Print matching results in the format: "filename:line_number:line_content"
+Results should be sorted first by filename, then by line number.
+Line numbers are 1-based.
+"""
+
+[[test_cases]]
+stdin = "error"
+expected_stdout = """app.txt:2:found an error here
+app.txt:4:another error occurred
+system.txt:1:error on startup"""
+setup_files = { "app.txt" = "all good\nfound an error here\nno problem\nanother error occurred", "system.txt" = "error on startup\nrunning fine\nshutdown", "data.csv" = "error,value\n1,2" }
+
+[[test_cases]]
+stdin = "hello"
+expected_stdout = """a.txt:1:hello world
+b.txt:2:say hello"""
+setup_files = { "a.txt" = "hello world\ngoodbye", "b.txt" = "greetings\nsay hello" }
+
+[[test_cases]]
+stdin = "notfound"
+expected_stdout = ""
+setup_files = { "test.txt" = "nothing matches here" }