[train] Increase default timeout for placement groups to 180s (NovaSky-AI#525)

# What does this PR do?

Increases the default timeout for placement groups to 3 minutes. I've
noticed that in multi-node clusters the initial worker startup time can
be easily ~ 1 min because it takes time to download all required
packages with uv. Thus, we need to account for additional time at
startup.

This could be increased even more, but this should work okay for small
clusters for now (the tradeoff is that you might wait too long even
though pg is not satisfiable)

---------

Signed-off-by: SumanthRH <sumanthrh@anyscale.com>

Author: SumanthRH

skyrl-train/docs/troubleshooting/troubleshooting.rst

@@ -6,7 +6,7 @@ Placement Group Timeouts
 -------------------------
 
 In SkyRL, we use Ray placement groups to request resources for different actors. In Ray clusters that autoscale with KubeRay, placement group creation can take a long time since the cluster might have to add a new node, pull the relevant image and start the container, etc. 
-You can use the ``SKYRL_RAY_PG_TIMEOUT_IN_S`` environment variable (Used in the ``.env`` file passed to the ``uv run`` command with ``--env-file``) to increase the timeout for placement group creation (By default, this is 60 seconds)
+You can use the ``SKYRL_RAY_PG_TIMEOUT_IN_S`` environment variable (Used in the ``.env`` file passed to the ``uv run`` command with ``--env-file``) to increase the timeout for placement group creation (By default, this is 180 seconds)
 
 Multi-node Training
 -------------------

skyrl-train/skyrl_train/utils/constants.py

@@ -1,7 +1,7 @@
 import os
 
 #
-SKYRL_RAY_PG_TIMEOUT_IN_S = int(os.environ.get("SKYRL_RAY_PG_TIMEOUT_IN_S", 60))
+SKYRL_RAY_PG_TIMEOUT_IN_S = int(os.environ.get("SKYRL_RAY_PG_TIMEOUT_IN_S", 180))
 """
 Timeout for allocating the placement group for different actors in SkyRL
 """