Investigate environment variable "MASTER_PORT" in spawn plugin

[awaelchli]

Proposed refactor

Investigate if setting the MASTER_PORT environment variable is needed here:
https://github.com/PyTorchLightning/pytorch-lightning/blob/4397f2d7481aad5d8a476e201d6a28daea5c4aa5/pytorch_lightning/strategies/launchers/spawn.py#L67

From comment by @carmocca

Motivation

We would like to rely on the cluster environment completely if possible, and not leak logic for specific variables.

Pitch

Remove and see if tests fail.

Additional context

---

If you enjoy Lightning, check out our other projects! ⚡

• Metrics: Machine learning metrics for distributed, scalable PyTorch applications.

• Lite: enables pure PyTorch users to scale their existing code on any kind of device while retaining full control over their own loops and optimization logic.

• Flash: The fastest way to get a Lightning baseline! A collection of tasks for fast prototyping, baselining, fine-tuning, and solving problems with deep learning.

• Bolts: Pretrained SOTA Deep Learning models, callbacks, and more for research and production with PyTorch Lightning and PyTorch.

• Lightning Transformers: Flexible interface for high-performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra.

cc @justusschock @awaelchli @rohitgr7

[awaelchli]

Just realized, it is there to fix the network port before spawning the processes. Otherwise, we would end up using a different port in each spawned process because of this:
https://github.com/PyTorchLightning/pytorch-lightning/blob/4dba492fb55e2862bb20f304e6bef43f1f151fc7/pytorch_lightning/plugins/environments/lightning_environment.py#L59

find_free_network_port() chooses a random free port. All processes would end up trying to connect on a different port. That's why we set the env variable BEFORE spawning processes so all of them end up using the same port.

[carmocca]

Can you add that as a comment in the PR? Thanks!