Process hangs in slot, failing to compute new incoming message

[jax-cn]

Env

1. edge branch
2. HB_PRINT=compute_short,push_short rebar3 as genesis_wasm shell
3. Operator: lpJ5Edz_8DbNnVDL0XdbsY9vCOs45NACzfI4jvo4Ba8
4. Process ID: ZIc9924GI_wMzPayOZAgVjaxasNq1rIQwdSZseGoh7M
5. Scheduler: GQ33BkPtZrqxA84vM8Zk-N2aO0toNNu_C-l-rawrBA (legacy SU)
6. halted slot: 1268

Bug

Normarlly, an incoming message will result in

pushed_message_to, process: Bf6JJ..hTRzo, slot: 15
starting_compute, proc_id: Bf6JJ..hTRzo, current: 15, target: 15
done, process: Bf6JJR2tl2Wr38O2-H6VctqtduxHgKF-NzRB9HhTRzo, slot: 15

But, for my process, an incoming message only result in

pushed_message_to, process: ZIc99..Goh7M, slot: 1298
starting_compute, proc_id: ZIc99..Goh7M, current: 1268, target: 1298

Expected results

starting_compute, proc_id: ZIc99..Goh7M, current: 1269, target: 1298
starting_compute, proc_id: ZIc99..Goh7M, current: 1270, target: 1298
...
starting_compute, proc_id: ZIc99..Goh7M, current: 1298, target: 1298
done, process: ZIc9924GI_wMzPayOZAgVjaxasNq1rIQwdSZseGoh7M, slot: 1298

Logs

When trying to reproduce the issue on another HyperBEAM Node, got this error

=ERROR REPORT==== 19-Aug-2025::10:14:03.535724 ===
Error in process <0.9374.0> with exit value:
{{nocatch,
     {necessary_message_not_found,<<"body/tags">>,
         <<"Link (to link): 1KXso6h8Hsi1FWXRYrR8sLU8kyBCSvavfFvdCyNM7j8">>}},
 [{hb_cache,report_ensure_loaded_not_found,3,
      [{file,"/root/HyperBEAM/src/hb_cache.erl"},{line,132}]},
  {hb_cache,ensure_all_loaded,3,
      [{file,"/root/HyperBEAM/src/hb_cache.erl"},{line,149}]},
  {maps,map_1,3,[{file,"maps.erl"},{line,942}]},
  {maps,map_1,3,[{file,"maps.erl"},{line,942}]},
  {maps,map,2,[{file,"maps.erl"},{line,927}]}, 
  {maps,map_1,3,[{file,"maps.erl"},{line,942}]},
  {maps,map_1,3,[{file,"maps.erl"},{line,942}]},
  {maps,map,2,[{file,"maps.erl"},{line,927}]}, 
  {dev_scheduler_cache,write,2,
      [{file,"/root/HyperBEAM/src/dev_scheduler_cache.erl"},{line,25}]},
  {lists,foreach_1,2,[{file,"lists.erl"},{line,2641}]},
  {dev_scheduler,'-cache_remote_schedule/2-fun-1-',2,
      [{file,"/root/HyperBEAM/src/dev_scheduler.erl"},{line,1238}]}]}

How to reproduce

1. Spawn a process on legacy (specify the scheduler to legacy SU) aos process_1
2. Reconnect to the HyperBEAM Node on edge branch aos process_1 --mainnet xxx
3. Write a handler that Send nested data structure to patch device each call.

Handlers.add("Cron", "Cron", function(msg) Send({ device = "patch@1.0", cache = {
	taskStats = {
	  succeeded = 59,
	  timestamp = os.time(),
	},
	tasks = {
	  ["sNWrdfUcR9kBpRPPPnJKFlel4j_z2rJ89PStNXITMto-1755453195384"] = {
		timestamp = os.time(),
		status = "succeeded",
	  }
	},
	timestamp = os.time(),
	workerTasks = {
	  timestamp = os.time(),
	}
} }) end)

5. Call the handler every 5 seconds.
6. After about 500 messages. Stop the Node.
7. Clear the cache-mainnet folder
8. Restart the Node.
9. Send another message to the process, and you might get the error log in a very high probability.

Is this a known issue?
I've met the issue a couple times. the previous process was D0na6AspYVzZnZNa7lQHnBt_J92EldK_oFtEPLjIexo

If need any future log or cache.tar to debug, please let me know.