drivers/sdcard: Do not release CS during the middle of read operations.

It seems that some cards do not tolerate releasing the card (by setting CS
high) after issuing CMD17 (and 18) and raising it again before reading
data. Somehow this causes the 0xfe data start marker not being read and
SDCard.readinto() is spinning forever (or until this byte is in the data).

This seems to fix weird behviour of SDCard.readblocks() returning different
data, also solved hanging os.mount() for my case with a 16GB Infineon card.

This stackexchange answer gives more context:
https://electronics.stackexchange.com/questions/307214/sd-card-spi-interface-issue-read-operation-returns-0x3f-0xff-instead-of-0x7f-0#307268

drivers/sdcard/sdcard.py

@@ -174,7 +174,7 @@ def readinto(self, buf):
         # read until start byte (0xff)
         while True:
             self.spi.readinto(self.tokenbuf, 0xff)
-            if self.tokenbuf[0] == 0xfe:
+            if self.tokenbuf[0] == _TOKEN_DATA:
                 break
 
         # read data
@@ -228,17 +228,22 @@ def readblocks(self, block_num, buf):
         assert nblocks and not len(buf) % 512, 'Buffer length is invalid'
         if nblocks == 1:
             # CMD17: set read address for single block
-            if self.cmd(17, block_num * self.cdv, 0) != 0:
+            if self.cmd(17, block_num * self.cdv, 0, release=False) != 0:
+                # release the card
+                self.cs(1)
                 raise OSError(5) # EIO
-            # receive the data
+            # receive the data and release card
             self.readinto(buf)
         else:
             # CMD18: set read address for multiple blocks
-            if self.cmd(18, block_num * self.cdv, 0) != 0:
+            if self.cmd(18, block_num * self.cdv, 0, release=False) != 0:
+                # release the card
+                self.cs(1)
                 raise OSError(5) # EIO
             offset = 0
             mv = memoryview(buf)
             while nblocks:
+                # receive the data and release card
                 self.readinto(mv[offset : offset + 512])
                 offset += 512
                 nblocks -= 1