dasLLAMA: PROBE_PATH's default was Windows-only

Copilot round 4, three threads on one issue. The write-cliff probe defaulted to `D:/_wcliff.bin`,
which on POSIX is not an absolute path at all — it is a relative `D:` directory that does not
exist, so the probe fails confusingly on every non-Windows host rather than writing anywhere.

Default is now cwd-relative `_wcliff.bin`, which resolves on every host; the registry entry and the
usage line say to point it at the drive under test, which is the whole point of the knob. Also
dropped `require strings`, unused since env_int64 replaced the to_int call.

Gates: compiles, lint 0 issues, test_env_registry 14/14, ENVIRONMENT.md regenerated.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ruKZPG8G7SPEUyZtvBzcT

Author: borisbat

modules/dasLLAMA/ENVIRONMENT.md

@@ -140,7 +140,7 @@ Apple Accelerate / AMX float lane. `DASLLAMA_ACCEL` arms the whole group.
 
 | Variable | Type | Default | Effect |
 |---|---|---|---|
-| `PROBE_PATH` | path | D:/_wcliff.bin | Scratch file the write-cliff probe writes; put it on the drive under test. |
+| `PROBE_PATH` | path | _wcliff.bin | Scratch file the write-cliff probe writes, cwd-relative by default; point it at the drive under test. |
 | `PROBE_GB` | number | 50 | Gigabytes the write-cliff probe writes before reporting. |
 | `DASLLAMA_BENCH_MODEL` | text | tinyllama | Model name for the isolated GEMM bench. |
 | `DASLLAMA_BENCH_NTOK` | number | unset | Token count for the isolated GEMM bench. |

modules/dasLLAMA/benchmarks/write_cliff_probe.das

@@ -6,10 +6,9 @@ options gen2
 // total. The .dlim writer is the caller that hits this: 75 GB at 275 MB/s where a 2.5 GB image
 // writes at ~1 GB/s.
 //   bin/daslang modules/dasLLAMA/benchmarks/write_cliff_probe.das
-//   PROBE_GB=50 PROBE_PATH=D:/_wcliff.bin bin/daslang .../write_cliff_probe.das
+//   PROBE_GB=50 PROBE_PATH=/mnt/nvme/_wcliff.bin bin/daslang .../write_cliff_probe.das
 
 require daslib/fio
-require strings
 require math
 require dasllama/dasllama_env
 
@@ -19,7 +18,8 @@ let WINDOW_CHUNKS = 32l   // one report line per 512 MB
 [export]
 def main() {
     let gb = max(env_int64("PROBE_GB", 50l), 1l)
-    let path = env_str("PROBE_PATH", "D:/_wcliff.bin")
+    // cwd-relative so the default works on every host; point it at the drive under test
+    let path = env_str("PROBE_PATH", "_wcliff.bin")
     let chunk_bytes = CHUNK_MB * 1024l * 1024l
     let chunks = max(gb * 1024l / CHUNK_MB, 1l)
     var buf : array<uint8>

modules/dasLLAMA/dasllama/dasllama_env.das

@@ -223,8 +223,8 @@ def private reg_vulkan(var o : array<EnvKnob>) {
       "Bucket gguf -> image conversion time by kind over the weight walk; one clock pair per tensor.")
     k(o, "DASLLAMA_ALLOW_INTERP_LOAD", EnvKind.flag, EnvArea.engine, "off",
       "Permit a big gguf load without -jit; the transforms run interpreted, so expect minutes per GB.")
-    k(o, "PROBE_PATH", EnvKind.path, EnvArea.benchmark, "D:/_wcliff.bin",
-      "Scratch file the write-cliff probe writes; put it on the drive under test.")
+    k(o, "PROBE_PATH", EnvKind.path, EnvArea.benchmark, "_wcliff.bin",
+      "Scratch file the write-cliff probe writes, cwd-relative by default; point it at the drive under test.")
     k(o, "PROBE_GB", EnvKind.number, EnvArea.benchmark, "50",
       "Gigabytes the write-cliff probe writes before reporting.")
     k(o, "DASLLAMA_MM_SMALLD", EnvKind.number, EnvArea.vulkan, "64",