src/sdcard/spi.rs: Wait for idle gap after escape

[fdomke]

After sending CMD55 (escape for ACMDs), the card is allowed to be busy again.

This fixes a regression that broke e.g. SanDisk Ultra 64GB SDXC cards.

[fdomke]

This regressed in d2334e7#diff-a73d1482336da8457a9bf37e49622a3850bd9fc5bc55af1162e1f2ba13a14a52L460-R469:

-    fn card_acmd(&mut self, command: u8, arg: u32) -> Result<u8, Error> {
-        self.card_command(CMD55, 0)?;
-        self.card_command(command, arg)    // << this used to wait for busy
+    fn card_acmd(&mut self, command: AcmdId, arg: u32) -> Result<u8, Error> {
+        self.card_command(super::CmdId::_55_AppCmd, 0)?;
+        self.card_acmd_after_escape(command, arg)
+    }
+
+    fn card_acmd_after_escape(&mut self, command: AcmdId, arg: u32) -> Result<u8, Error> {
+        let mut buf = [
+            0x40 | command as u8,
+            (arg >> 24) as u8,
+            (arg >> 16) as u8,
+            (arg >> 8) as u8,
+            arg as u8,
+            0,
+        ];
+        buf[5] = crc7(&buf[0..5]);
+
+        self.write_bytes(&buf)?;  // << this does not wait anymore

[...]
    fn card_command(&mut self, command: CmdId, arg: u32) -> Result<u8, Error> {
        if command != super::CmdId::_0_GoIdleState && command != super::CmdId::_12_StopTransmission
        {
             self.wait_not_busy(Delay::new_command())?;
         }
[...]

after which wait_not_busy was not anymore called after CMD55.

[thejpster]

Oops. Thanks for the fix.