feat: add retry with exponential backoff for transient registry errors

Add tenacity-based retry logic to get_rpmdb() and inspect_image_ref()
so that transient HTTP errors from registries (502, 503, 429, connection
resets, DNS failures) are retried automatically instead of failing the
pipeline.

- Retry up to 4 attempts with exponential backoff (5-60s) and jitter
- Only retry on transient errors; permanent errors fail immediately
- Log retry attempts to stderr for visibility in Tekton task logs
- Add tenacity dependency to pyproject.toml and Pipfile
- Add comprehensive tests for retry behavior and error classification
- Update .gitignore for __pycache__ and uv.lock

Assisted-by: Claude Code (Claude Opus 4.6)
Signed-off-by: Ernesto González <ergonzal@redhat.com>

Author: ernesgonzalez33

.gitignore

@@ -1 +1,3 @@
 *.egg-info
+__pycache__/
+uv.lock

Pipfile

@@ -21,6 +21,7 @@ pylint-pytest = "*"
 authlib = "*"
 kubernetes = "*"
 pytest-cov = "*"
+tenacity = "*"
 
 [dev-packages]
 pytest = "*"

Pipfile.lock

@@ -16,6 +16,7 @@ dependencies = [
     "pyyaml",
     "odcs[client]",
     "kubernetes",
+    "tenacity",
 ]
 
 [project.scripts]
@@ -36,3 +37,5 @@ target-version = ["py311"]
 
 [tool.pylint]
 load-plugins = ["pylint_pytest"]
+
+