[FLINK-37138][sql-gateway] Fix failed SqlYARNApplicationITCase in hadoop3 profile

[fsk119]

What is the purpose of the change

YARN deployment doesn't the directories as flink-dist does. Actually, it uses the following structures

.
└── ohmeatball
    ├── appcache
    │   └── application_1737534562065_0001
    │       ├── container_1737534562065_0001_01_000001
    │       │   ├── lib
    │       │   ├── opt
    │       │   ├── plugins
    │       │   │   ├── external-resource-gpu
    │       │   │   ├── metrics-datadog
    │       │   │   ├── metrics-graphite
    │       │   │   ├── metrics-influx
    │       │   │   ├── metrics-jmx
    │       │   │   ├── metrics-otel
    │       │   │   ├── metrics-prometheus
    │       │   │   ├── metrics-slf4j
    │       │   │   └── metrics-statsd
    │       │   └── tmp
    │       └── filecache
    │           ├── 10
    └── filecache

Actually the config.yaml is at the working directory(container_1737534562065_0001_01_000001). So sql-gateway should not load the conf from the $FLINK_CONF_DIR. Considering YarnApplicationClusterEntryPoint or KubernetesApplicationClusterEntrypoint boths read the conf and put its content in the StreamExecutionEnvironment,
we just get the conf from StreamExecutionEnvironment.

[fsk119]

@flinkbot run azure

[flinkbot]

CI report:

• e616aef UNKNOWN

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build