Call MinimumInstanceChecker when an agent is bound

[jglick]

Extracted from #1121.

I was testing a cloud config like this

amazonEC2:
  name: …
  region: …
  cleanUpOrphanedNodes: true
  noDelayProvisioning: true
  sshKeysCredentialsId: …
  templates:
  - ami: …
    amiType:
      unixData:
        sshPort: "22"
    associatePublicIp: false
    avoidUsingOrphanedNodes: true
    connectBySSHProcess: false
    connectionStrategy: PRIVATE_IP
    deleteRootOnTermination: false
    description: …
    ebsEncryptRootVolume: DEFAULT
    ebsOptimized: false
    enclaveEnabled: false
    hostKeyVerificationStrategy: 'OFF'
    idleTerminationMinutes: "0" # ℹ️
    javaPath: java
    labelString: …
    maxTotalUses: 1 # ℹ️
    metadataEndpointEnabled: true
    metadataHopsLimit: 1
    metadataSupported: true
    metadataTokensRequired: false
    minimumNumberOfInstances: 0
    minimumNumberOfSpareInstances: 2 # ℹ️
    mode: EXCLUSIVE
    monitoring: false
    numExecutors: 1 # ℹ️
    remoteAdmin: …
    remoteFS: …
    securityGroups: …
    stopOnTerminate: false
    subnetId: …
    t2Unlimited: false
    tenancy: Default
    type: …
    useEphemeralDevices: false
  useInstanceProfileForCredentials: false

The intention is to keep two instances booted, connected, and ready to use at all times, treating them as “one-shot” (destroy after use, so there is no chance of cross-build state contamination). However after running a build, which consumed one instance, it was not replenished with a fresh one, until a periodic task ran every ten minutes. This is not nearly fast enough to keep up with even a moderately frequent build schedule. With this patch, the moment an instance is assigned to a build, the process of launching a fresh instance is started, so that the pool remains sufficiently warm so long as you do not trigger too many builds for the pool size to keep up.

There may well be other situations in which instances need to be refreshed; this is just the one I found in this scenario.

[jglick]

https://github.com/jenkinsci/ec2-plugin/pull/1124/checks?check_run_id=47157908661

org.opentest4j.AssertionFailedError: expected: <2> but was: <3>
	at org.junit.jupiter.api.AssertionFailureBuilder.build(AssertionFailureBuilder.java:151)
	at org.junit.jupiter.api.AssertionFailureBuilder.buildAndThrow(AssertionFailureBuilder.java:132)
	at org.junit.jupiter.api.AssertEquals.failNotEqual(AssertEquals.java:197)
	at org.junit.jupiter.api.AssertEquals.assertEquals(AssertEquals.java:150)
	at org.junit.jupiter.api.AssertEquals.assertEquals(AssertEquals.java:145)
	at org.junit.jupiter.api.Assertions.assertEquals(Assertions.java:531)
	at hudson.plugins.ec2.EC2RetentionStrategyTest.testRetentionDespiteIdleWithMinimumInstanceActiveTimeRange(EC2RetentionStrategyTest.java:928)

According to CloudBees PCT testing this test is known to be flaky. Tried it locally in this branch and it passed 5×, so I am guessing the failure is unrelated to my changes.