chore(deps): bump agentfield to >=0.1.73 (#56)

Pulls in the OpenCodeProvider stderr-error-detection fix released in
agentfield 0.1.73 (Agent-Field/agentfield#525). Without this bump the
Docker image keeps resolving to the cached older agentfield, so the
companion SWE-AF model-prefix fix from #55 lands but the harness still
silently swallows opencode "Model not found" / auth errors at exit 0.

Updates the pin in:
- requirements-docker.txt (Docker build path)
- requirements.txt (bare-metal install path)
- pyproject.toml (was lagging at 0.1.9)

Verified agentfield-0.1.73-py3-none-any.whl ships
_extract_opencode_error and _OPENCODE_STDERR_ERROR_PATTERNS.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: AbirAbbas

pyproject.toml

@@ -4,7 +4,7 @@ version = "0.1.0"
 description = "Autonomous SWE agent node for AgentField"
 requires-python = ">=3.12"
 dependencies = [
-    "agentfield>=0.1.9",
+    "agentfield>=0.1.73",
     "pydantic>=2.0",
     # Compatibility pin: newer SDK builds have surfaced
     # "Unknown message type: rate_limit_event" during streaming.

requirements-docker.txt

@@ -2,6 +2,6 @@
 #
 # Same runtime dependencies as requirements.txt.
 
-agentfield>=0.1.67
+agentfield>=0.1.73
 pydantic>=2.0
 claude-agent-sdk==0.1.20

requirements.txt

@@ -2,6 +2,6 @@
 #
 # Install: python -m pip install -r requirements.txt
 
-agentfield>=0.1.67
+agentfield>=0.1.73
 pydantic>=2.0
 claude-agent-sdk==0.1.20