50fa8744fd
Delete All worked but reset the preset counter: the handler never re-read the preset (unlike ramExecute), saved N+1, then re-ran the Explore with the OLD sp (counter N). Every page answered cntr_low, nothing was accepted, and ramExplore's final ramSaveCntr wrote N back over the preset. - ramDeleteFromExplorer re-reads the preset first (spRefreshFromPreset 'ram-card-sel'), advances and saves the counter only for a packet the card accepted (spPorAccepted) - a rejected packet leaves it untouched - and continues the follow-up Explore from the consumed counter instead of replaying it. - the delete result also checks the remote command's SW via a new `ramRemoteSwOk` (9000 / 61xx / 62xx / 63xx / CAFE, mirroring the server's `_ram_remote_sw_ok`): a por_ok packet whose DELETE was refused (e.g. 6A88) no longer shows "OK" (the counter still advances, the card consumed it). - ramExplore only saves the counter when it advanced, so a stale caller can never lower a preset counter again. - tests: flow tests for accepted / rejected / refused-remote-SW deletes (refresh order, saved counter, the counter passed to the re-explore) and the ramRemoteSwOk set. 615 frontend / 495 Python green; version 3.6.7; sw simple-v280.