Fix language instruction for OpenFaucetTeleop and XMovePickTeleop to match training data

The policy instruction comes from environment_config.dr_state_dict.language, which held strings that differ from the training data:
- OpenFaucetTeleop: 'move forward to the faucet and open it' -> 'turn the faucet'
- XMovePickTeleop: 'bend to pick up the {}' -> 'move forward to pick up the {}'
Also aligned dr_cfgs.language.instructions, episode 'tasks' and meta/tasks.jsonl. No other fields changed.

G1WholebodyOpenFaucetTeleop-v0/dr-level-0/meta/tasks.jsonl

@@ -1 +1 @@
-{"task_index": 0, "task": "move forward to the faucet and open it"}
+{"task_index": 0, "task": "turn the faucet"}

G1WholebodyOpenFaucetTeleop-v0/dr-level-1/meta/tasks.jsonl

@@ -1 +1 @@
-{"task_index": 0, "task": "move forward to the faucet and open it"}
+{"task_index": 0, "task": "turn the faucet"}

G1WholebodyOpenFaucetTeleop-v0/dr-level-2/meta/tasks.jsonl

@@ -1 +1 @@
-{"task_index": 0, "task": "move forward to the faucet and open it"}
+{"task_index": 0, "task": "turn the faucet"}

G1WholebodyOpenFaucetTeleop-v0/dr-level-hard/meta/tasks.jsonl

@@ -1 +1 @@
-{"task_index":0,"task":"turn the faucet"}
+{"task_index": 0, "task": "turn the faucet"}

G1WholebodyXMovePickTeleop-v0/dr-level-hard/meta/tasks.jsonl

@@ -1 +1 @@
-{"task_index":0,"task":"move forward to pick up the cracker box"}
+{"task_index": 0, "task": "move forward to pick up the cracker box"}