layered: actually get slice_ns and max_exec_ns from CO:RE

Layered previously hardcoded the defaults for slice_ns and max_exec_us. This
was changed in a recent PR to get these from CO:RE in the running kernel. This
change doesn't work because the old values are still set a few lines below.

Remove the old method of setting these values allowing CO:RE to win.

Test plan:
- CI

scheds/rust/scx_layered/src/main.rs

@@ -1552,8 +1552,6 @@ impl<'a> Scheduler<'a> {
         skel_builder.obj_builder.debug(opts.verbose > 1);
         init_libbpf_logging(None);
         let mut skel = scx_ops_open!(skel_builder, open_object, layered)?;
-        skel.maps.rodata_data.slice_ns = scx_enums.SCX_SLICE_DFL;
-        skel.maps.rodata_data.max_exec_ns = 20 * scx_enums.SCX_SLICE_DFL;
 
         // Initialize skel according to @opts.
         skel.struct_ops.layered_mut().exit_dump_len = opts.exit_dump_len;
@@ -1562,12 +1560,8 @@ impl<'a> Scheduler<'a> {
         // Running scx_layered inside a PID namespace would break the
         // following.
         skel.maps.rodata_data.layered_tgid = std::process::id() as i32;
-        skel.maps.rodata_data.slice_ns = opts.slice_us * 1000;
-        skel.maps.rodata_data.max_exec_ns = if opts.max_exec_us > 0 {
-            opts.max_exec_us * 1000
-        } else {
-            opts.slice_us * 1000 * 20
-        };
+        skel.maps.rodata_data.slice_ns = scx_enums.SCX_SLICE_DFL;
+        skel.maps.rodata_data.max_exec_ns = 20 * scx_enums.SCX_SLICE_DFL;
         skel.maps.rodata_data.nr_possible_cpus = *NR_CPUS_POSSIBLE as u32;
         skel.maps.rodata_data.smt_enabled = cpu_pool.nr_cpus > cpu_pool.nr_cores;
         skel.maps.rodata_data.has_little_cores = topo.has_little_cores();