osdc: restart read on truncate/discard

osdc: restart read on truncate/discard

This can be reliably reproduced by reading a large file (~500MB) on a
ceph-fuse mount and then evicting the client.

Now we see in the logs:

	2020-10-21T01:46:29.679+0000 7f00beffd700 10 objectcacher.object(10000000000.00000030/head) truncate object[10000000000.00000030/head oset 0x7f00a0006b08 wr 0/0] to 0
	2020-10-21T01:46:29.679+0000 7f00beffd700 10 objectcacher.object(10000000000.00000030/head) restarting reads post-truncate
	...
	2020-10-21T01:46:29.762+0000 7f00bffff700 20 objectcacher.object(10000000000.00000030/head) map_read error bh[ 0x7f0070004e30 1048576~3145728 0x7f008404a720 (0) v 0 error=-108] waiters = {}
	2020-10-21T01:46:29.762+0000 7f00bffff700 10 objectcacher readx hit bh bh[ 0x7f0070004e30 1048576~3145728 0x7f008404a720 (0) v 0 error=-108] waiters = {}
	2020-10-21T01:46:29.762+0000 7f00bffff700 10 objectcacher readx has all buffers
	2020-10-21T01:46:29.762+0000 7f00bffff700 20 objectcacher readx done 0x7f007c00a9b0 -108
	...
	2020-10-21T01:46:29.763+0000 7f0077fff700  3 client.4497 ll_read 0x7f0080023720 202637312~131072 = -108

Fixes: https://tracker.ceph.com/issues/46434
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
(cherry picked from commit b53390031378a06980fa8e7dae074a7095307c12)

Conflicts:
	src/osdc/ObjectCacher.cc
	  - keep the specific data type instead using auto