secret_detection job in 14.9 fails in small repositories: "fatal: --unshallow on a complete repository does not make sense"

Summary

Steps to reproduce

• Have a project with a small number of commits, e.g. 7
• Include this in your CI:

include:
  - template: Security/Secret-Detection.gitlab-ci.yml

secret_detection:
  variables:
    # Run on the entire Git history of a repository.
    # https://docs.gitlab.com/ee/user/application_security/secret_detection/#custom-settings-example
    SECRET_DETECTION_HISTORIC_SCAN: "true"

• Run a CI pipeline

What is the current bug behavior?

Executing "step_script" stage of the job script 00:01
Using docker image sha256:f79526522cc5925f34865f06a3564f32729f77040173a1abc939680c4b0b6995 for registry.gitlab.com/security-products/secrets:3 with digest registry.gitlab.com/gitlab-org/security-products/analyzers/secrets@sha256:d87684cc7e034c583b1582069d6f6f845a9276f562ae385428b4c8508334ac3a ...
$ if [ -n "$CI_COMMIT_TAG" ]; then echo "Skipping Secret Detection for tags. No code changes have occurred."; exit 0; fi
$ if [ "$SECRET_DETECTION_HISTORIC_SCAN" == "true" ] # collapsed multi-line command
historic scan
fatal: --unshallow on a complete repository does not make sense
Uploading artifacts for failed job

Results of GitLab environment info

Just updated from 14.8.2 to 14.9.2

Bug location

ef2904cd for !80326 (merged)

cc @marcel.amirault @zrice @matteeyah

This might be relevant: https://github.com/jderusse/docker-gitsplit/issues/18

Workaround

Disable SECRET_DETECTION_HISTORIC_SCAN.

(~regression:14.9)