Pin mlflow==1.26.1 in credit_defaults_model to fix ImageBuildFailure

The unpinned mlflow dependency resolves to a version (>=2.18.0) that
requires Python >=3.9, which conflicts with the model's Python 3.8.15
environment. This causes the deployment Docker image build to fail.

Pin mlflow to 1.26.1 (the version that originally created the model)
in both conda.yaml and requirements.txt.

Author: Chakradhar886

tutorials/get-started-notebooks/deploy/credit_defaults_model/conda.yaml

@@ -4,7 +4,7 @@ dependencies:
 - python=3.8.15
 - pip<=21.2.4
 - pip:
-  - mlflow
+  - mlflow==1.26.1
   - cloudpickle==2.2.0
   - psutil==5.8.0
   - scikit-learn==0.24.2

tutorials/get-started-notebooks/deploy/credit_defaults_model/requirements.txt

@@ -1,4 +1,4 @@
-mlflow
+mlflow==1.26.1
 cloudpickle==2.2.0
 psutil==5.8.0
 scikit-learn==0.24.2