According to the SolarWinds NPM Administrator Guide, you can edit the node properties of any device and change the polling method to suit your needs. The polling method determines what data is collected and displayed for the node. The "Most Devices: SNMP and ICMP" polling method collects and displays the most comprehensive data, including CPU load, memory utilization, interface statistics, and more. The "Status Only:
ICMP" polling method only collects and displays the node status, average response time, and packet loss. If you want to monitor the other statistics on switch A, you should change the polling method to "Most Devices:
SNMP and ICMP". This will not affect the historical data that has already been collected for switch A, as it will be stored in the database. However, deleting switch A and adding it again, or adding switch A again without deleting it first, will result in losing the historical data or creating duplicate nodes. References: SolarWinds NPM Administrator Guide, page 78-79; Polling methods used by Orion

Universal Device Pollers (UnDPs) are custom pollers that allow you to monitor specific aspects of your network devices using SNMP. You can apply UnDPs to any object type that supports SNMP polling, such as interfaces, nodes, and volumes. You cannot apply UnDPs to applications, because they are not SNMP-based objects. Applications are monitored using Application Performance Monitor (APM), which uses different methods, such as WMI, scripts, or agents, to collect data. References: Monitor custom statistics based on OIDs with Universal Device Pollers in NPM, Assign Universal Device Pollers to devices monitored in NPM, Add and assign Universal Device Pollers

For any node you monitor, you cannot use Network Discovery to automatically add child volumes or interfaces for monitoring. Network Discovery is a tool that helps you identify and add new devices to your network, but it does not automatically monitor all the components of the discovered devices. You need to manually select the volumes or interfaces you want to monitor from the list of resources on each node. You can also use the Network Sonar Results Wizard to filter and add the resources you want to monitor after running a discovery job. References:
Discover your network with NPM - SolarWinds (see the Monitoring Settings panel section) Network Discovery Tool - Automated Discovery Software | SolarWinds (see the Monitor Devices section)

SolarWinds NPM can monitor various metrics on an Ethernet switch, including configuration changes, duplex mismatches, and CPU utilization. Configuration changes are tracked by the Network Configuration Manager (NCM) integration, which allows you to view, compare, and restore configuration files, as well as receive alerts on configuration changes1. Duplex mismatches are detected by the topology polling feature, which collects and displays the duplex mode of the connected devices and interfaces, and alerts you when there is a mismatch2. CPU utilization is monitored by the SNMP polling feature, which collects and displays the CPU load of the switch, as well as other performance metrics such as memory utilization, interface statistics, and more3. Buffer misses are not monitored by NPM, as they are not part of the standard SNMP MIBs that NPM supports. References: Monitor configuration changes on network devices with NPM; Detect and predict duplex mismatches in NPM; Monitor CPU load, memory utilization, and buffer usage on switches and routers;
[Buffer Misses - Forum].
SolarWinds Network Performance Monitor (NPM) is a comprehensive monitoring solution designed to provide a wide range of metrics for network devices. For an Ethernet switch, NPM can monitor configuration changes, buffer misses, duplex mismatches, and CPU utilization1