docs: Colab notebook for full GPU pipeline run

End-to-end Colab notebook (notebooks/02_colab_run.ipynb) that produces a
real cohort of PD-L1 binder designs with measured metrics. Workflow:

  1. Install RFdiffusion + ProteinMPNN + ColabFold + BinderDiffuser
  2. Download PDB 5O45, trim to chain B (PD-L1 IgV)
  3. Generate 10 motif-scaffolded backbones with RFdiffusion
  4. Design 4 sequences per backbone with ProteinMPNN
  5. Re-fold each sequence with ColabFold (AF2 multimer)
  6. Compute scRMSD, scTM, pLDDT, ipTM, pAE_interface
  7. Filter, rank, render scatter and violin plots
  8. Tarball and download artifacts

Smoke-test parameters chosen to fit Colab's free T4 in under 30 minutes.
Bump NUM_BACKBONES and SEQS_PER_BACKBONE for a larger run.

Adds Colab launch badge, CI status badge, and license badge to README.

Author: deepmind11

README.md

@@ -2,6 +2,10 @@
 
 **De novo protein binder design via motif-scaffolded diffusion + sequence design + structural validation.**
 
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/deepmind11/BinderDiffuser/blob/main/notebooks/02_colab_run.ipynb)
+[![CI](https://github.com/deepmind11/BinderDiffuser/actions/workflows/ci.yml/badge.svg)](https://github.com/deepmind11/BinderDiffuser/actions/workflows/ci.yml)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+
 ![Hero](figures/hero.png)
 
 BinderDiffuser is an end-to-end pipeline for designing protein binders against