juju-mongodb process hangs around after package removal

Steps to reproduce:

Running on Trusty with juju versions:

ii juju-core 1.18.0-0ubuntu1
ii juju-local 1.18.0-0ubuntu1
ii juju-mongodb 2.4.9-0ubuntu3

With environments.yaml as:

default: local

environments:
  local:
    type: local
    container: kvm

juju status output (doesn't matter what charms are deployed or even if a machine is added, showing for completeness):

environment: local
machines:
  "0":
    agent-state: started
    agent-version: 1.18.0.1
    dns-name: localhost
    instance-id: localhost
    series: trusty
  "1":
    agent-state: started
    agent-version: 1.18.0.1
    dns-name: 10.0.3.192
    instance-id: poe-local-machine-1
    series: precise
    hardware: arch=amd64 cpu-cores=1 mem=1024M root-disk=8192M
services:
  glance:
    charm: cs:precise/glance-30
    exposed: false
    relations:
      cluster:
      - glance
      identity-service:
      - keystone
      image-service:
      - nova-cloud-controller
      shared-db:
      - mysql
    units:
      glance/0:
        agent-state: pending
        agent-version: 1.18.0.1
        machine: "1"
        public-address: 10.0.3.192
  keystone:
    charm: cs:precise/keystone-31
    exposed: false
    relations:
      cluster:
      - keystone
      identity-service:
      - glance
      - nova-cloud-controller
      - openstack-dashboard
      shared-db:
      - mysql
    units:
      keystone/0:
        agent-state: pending
        agent-version: 1.18.0.1
        machine: "1"
        public-address: 10.0.3.192
  mysql:
    charm: cs:precise/mysql-39
    exposed: false
    relations:
      cluster:
      - mysql
      shared-db:
      - glance
      - keystone
      - nova-cloud-controller
    units:
      mysql/0:
        agent-state: started
        agent-version: 1.18.0.1
        machine: "1"
        public-address: 10.0.3.192
  nova-cloud-controller:
    charm: cs:precise/nova-cloud-controller-32
    exposed: false
    relations:
      amqp:
      - rabbitmq-server
      cluster:
      - nova-cloud-controller
      identity-service:
      - keystone
      image-service:
      - glance
      shared-db:
      - mysql
    units:
      nova-cloud-controller/0:
        agent-state: pending
        agent-version: 1.18.0.1
        machine: "1"
        public-address: 10.0.3.192
  openstack-dashboard:
    charm: cs:precise/openstack-dashboard-15
    exposed: false
    relations:
      cluster:
      - openstack-dashboard
      identity-service:
      - keystone
    units:
      openstack-dashboard/0:
        agent-state: pending
        agent-version: 1.18.0.1
        machine: "1"
        public-address: 10.0.3.192
  rabbitmq-server:
    charm: cs:precise/rabbitmq-server-21
    exposed: false
    relations:
      amqp:
      - nova-cloud-controller
      cluster:
      - rabbitmq-server
    units:
      rabbitmq-server/0:
        agent-state: pending
        agent-version: 1.18.0.1
        machine: "1"
        public-address: 10.0.3.192

Output of juju version of mongodb:

root 27184 1 1 20:53 ? 00:00:02 /usr/lib/juju/bin/mongod --auth --dbpath=/home/poe/.juju/local/db --sslOnNormalPorts --sslPEMKeyFile /home/poe/.juju/local/server.pem --sslPEMKeyPassword xxxxxxx --bind_ip 0.0.0.0 --port 37017 --noprealloc --syslog --smallfiles
poe 28089 13413 0 20:57 pts/0 00:00:00 grep --color=auto mongo

Destroy the environment:

juju destroy-environment local

Remove packages:

sudo apt-get remove juju-local juju-mongodb juju-core

Check if the /usr/lib/juju/bin/mongod process is still there (it is)

Attempt to re-bootstrap a new local environment

$ juju bootstrap
ERROR cannot use 37017 as state port, already in use

Kill the process (sometimes it even attempts to respawn). Make sure the process is finally gone and re-bootstrap to a working environment.