Fix parallel code gen in RT backend

hercules_cg/src/rt.rs

hercules_opt/src/inline.rs

hercules_rt/src/lib.rs

juno_samples/fork_join_tests/src/cpu.sch

juno_samples/fork_join_tests/src/fork_join_tests.jn

juno_samples/fork_join_tests/src/gpu.sch