fix async self-update, check user permissions on login

[Panaetius]

closes #65
also sets default job timeout to 10 hours as some systems don't allow high values like 24 hours

[github-actions]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Handle API client creation errors gracefully The unwrap() call on CscsApi::new() can panic if the platform is invalid; replace it with proper error handling to make the login process more robust. coman/src/cscs/handlers.rs [88-105] let token = client_credentials_login(client_id_secret, client_secret_secret).await?; // figure out what platform the user has access to and set it in config let mut config = Config::new()?; let source = config.value_source("cscs.current_platform")?; if source.1 || source.2 { // don't override setting if it's already provided return Ok(()); } for platform in ComputePlatform::iter() { - let api_client = CscsApi::new(token.0.0.clone(), Some(platform.clone())).unwrap(); - if (api_client.list_systems().await).is_ok() { - config.set("cscs.current_platform", platform.to_string(), true)?; - break; + match CscsApi::new(token.0.0.clone(), Some(platform.clone())) { + Ok(api_client) => { + if api_client.list_systems().await.is_ok() { + config.set("cscs.current_platform", platform.to_string(), true)?; + break; + } + } + Err(_) => continue, // Skip invalid platforms instead of panicking } } Ok(()) Suggestion importance[1-10]: 8 __ Why: The suggestion correctly identifies a potential panic from unwrap() on CscsApi::new() and provides a better approach using match to handle the error gracefully. This improves robustness by preventing crashes due to invalid platform configurations.	Medium
	Improve error handling in async update call The update function is called within a blocking task, but the error handling could be improved to ensure proper propagation of errors from the update operation. coman/src/cli/app.rs [607-608] -update()?; +let result = tokio::task::spawn_blocking(update).await??; std::fs::write(&stamp_path, Local::now().to_rfc3339())?; +result Suggestion importance[1-10]: 3 __ Why: The suggestion proposes changing the error handling for the update function call, but the actual change doesn't significantly improve error propagation since tokio::task::spawn_blocking(update).await?? already handles errors properly. The suggested reassignment of result is unnecessary and doesn't add value.	Low
General	Improve configuration access error handling The use of unwrap_or_default() may hide potential configuration issues; consider returning a more explicit error when accessing nested configuration layers fails. coman/src/config.rs [335-345] // Returns tuple of bool saying whether a values is set in (default, global, project local) config pub fn value_source(&self, key_path: &str) -> Result<(bool, bool, bool)> { Ok(( self.default_layer.get(key_path).is_some(), - self.global_layer.get(key_path).unwrap_or_default().is_some(), + self.global_layer.get(key_path).is_some(), self.project_layer .as_ref() - .map(|l| l.get(key_path).unwrap_or_default().is_some()) + .map(|l| l.get(key_path).is_some()) .unwrap_or(false), )) } Suggestion importance[1-10]: 5 __ Why: The suggestion addresses a potential issue with unwrap_or_default() in configuration access, which could mask errors. However, the improvement is moderate because the current implementation likely works correctly in practice, and the change only affects error reporting rather than core functionality.	Low