Skipping LFS pull not possible after upgrading to gitlab-runner 11.9

Summary

Running a CI job in a repo with LFS pulls all LFS files even if GIT_LFS_SKIP_SMUDGE: 1 is defined in gitlab-ci.yml

I downgraded to 11.8 to get back to the expected behaviour.

Steps to reproduce

Given a repo with LFS files and a gitlab-ci.file:

stages:
  - check
a_check_job:
  stage: check
  variables:
    GIT_STRATEGY: clone
    GIT_LFS_SKIP_SMUDGE: 1
  script:
    - git fetch origin
    - echo "This job should be quick. No pulling of LFS files"

Actual behavior

All the LFS are pulled down to working dir. This takes a long time (hours) and might fill the disk (depending on size, of course)

Expected behavior

Quick checkout of working folder and start of the script block.

Relevant logs and/or screenshots

Raw output from job log:

[0KRunning with gitlab-runner 11.9.0 (692ae235)
[0;m[0K  on tomato b6861495
[0;m[0KUsing Shell executor...
[0;msection_start:1553425439:prepare_script
[0KRunning on tomato.uio.no...
section_end:1553425439:prepare_script
[0Ksection_start:1553425439:get_sources
[0K[0;33mDEPRECATION: this GitLab server doesn't support refspecs, gitlab-runner 12.0 will no longer work with this version of GitLab[0;m
[32;1mCloning repository...[0;m
Cloning into '/builder/b6861495/0/apps/vcpipe-bundle'...
[32;1mChecking out 7bc71163 as dev...[0;m
Downloading LFS objects:   0% (0/1), 0 B | 0 B/s                                
Downloading LFS objects:   0% (0/339), 0 B | 0 B/s                              
Downloading LFS objects:   0% (0/339), 2.2 MB | 0 B/s         

Environment description

Gitlab Omnibus 11.8.3 on our own server (RHEL7) and gitlab-runner 11.9 shared runner (also RHEL7). Shell executor.

Used GitLab Runner version

gitlab-runner 11.9.0

(output of gitlab-runner --version is not available, as I downgraded to 11.8)