Trim internal process notes from the card; the data facts are unchanged

README.md

@@ -17,8 +17,8 @@ Code and analysis: **[github.com/twu3202/EEG_SSVEP_Drone](https://github.com/twu
 Motor-imagery data from the same cap: **[Twu31/cap32-mi-eeg](https://huggingface.co/datasets/Twu31/cap32-mi-eeg)**
 
 > ⚠ **These files hold 10 posterior channels, not 32.** Channel selection was applied at
-> record time — an irreversible mistake that means these sessions cannot be re-analysed with
-> a different montage. `meta_json.channels_full` lists the 32 the cap was actually wearing.
+> record time, so these sessions cannot be re-analysed with a different montage.
+> `meta_json.channels_full` lists the 32 the cap was actually wearing.
 
 ## Quick start
 
@@ -85,7 +85,7 @@ not a property of the subject.
 
 ## Caveats
 
-1. **10 channels only** — see the warning above. Irreversible.
+1. **10 channels only** — see the warning above.
 2. **Dry electrodes over the occiput seat poorly.** SSVEP lives on Oz/O1/O2/POz; this is the
    hardest region for a dry cap, and it bounds everything here.
 3. **Timing is critical.** At 12 Hz one cycle is 83 ms, so a 4 ms epoch-boundary error is 14°