[Fleet] Surface conditions making the agent "not upgradeable" on "error upgrading agent" message

[criamico]

When upgrading an agent, we read the local_metadata.elastic.agent.upgradeable state from the elastic agent data, but we also do additional checks to mark the agent as upgradeable:

kibana/x-pack/plugins/fleet/common/services/is_agent_upgradeable.ts

Lines 16 to 43 in 3ec310b

	export function isAgentUpgradeable(
	agent: Agent,
	latestAgentVersion: string,
	versionToUpgrade?: string
	) {
	let agentVersion: string;
	if (typeof agent?.local_metadata?.elastic?.agent?.version === 'string') {
	agentVersion = agent.local_metadata.elastic.agent.version;
	} else {
	return false;
	}
	if (agent.unenrollment_started_at || agent.unenrolled_at) {
	return false;
	}
	if (!agent.local_metadata.elastic.agent.upgradeable) {
	return false;
	}
	if (isAgentUpgrading(agent)) {
	return false;
	}
	if (getRecentUpgradeInfoForAgent(agent).hasBeenUpgradedRecently) {
	return false;
	}
	if (versionToUpgrade !== undefined) {
	return isNotDowngrade(agentVersion, versionToUpgrade);
	}
	return isAgentVersionLessThanLatest(agentVersion, latestAgentVersion);
	}

In some cases this can result in this error message, that gives no explanations about why the agent cannot be upgraded:

Add additional information to the error message explaining why the agent is not upgradeable. We can easily get this info from is_agent_upgradeable itself.

Example

One example is the following agent, that has "upgradeable": true, from the agent itself but "upgrade_started_at": "2023-09-12T21:01:57.796Z" and "upgraded_at": null that mark the agent as already upgrading and so don't allow starting another upgrade.

{
  "item": {
    "id": "078b8cfa-fd39-461d-85a7-cee57d841204",
    "type": "PERMANENT",
    "active": true,
    "enrolled_at": "2023-03-03T11:02:58Z",
    "upgraded_at": null,
    "upgrade_started_at": "2023-09-12T21:01:57.796Z",
    "policy_id": "019cf5a",
    "last_checkin": "2023-11-22T20:29:12Z",
    "last_checkin_status": "online",
    "last_checkin_message": "Running",
    "policy_revision": 279,
    "agent": {
      "id": "078b",
      "version": "8.6.2"
    },
    "local_metadata": {
      "elastic": {
        "agent": {
          "build.original": "8.6.2 (build: 913c02bea9b13dec4d5c5f3057b5b397344e3298 at 2023-02-13 16:51:46 +0000 UTC)",
          "id": "078b8cfa-fd39-461d-85a7-cee57d841204",
          "log_level": "info",
          "snapshot": false,
          "upgradeable": true,
          "version": "8.6.2"
        }
      },
    "status": "offline"
  }

In this case the error message should say something like

agent N.. is not upgradeable: an upgrade attempt started at "2023-09-12T21:01:57.796Z" and was not completed

[elasticmachine]

Pinging @elastic/fleet (Team:Fleet)

[kpollich]

Related: #173281