tog: uppercase alpha key maps that open new views ok stsp@

tog: reset view->offset when clearing log view state Don't move the selection cursor due to a bogus offset after closing child views following use of 'B', C-l, and backspace log view key maps. ok stsp@

tog: use view->resize idiom to ensure log is populated Replace view->mode == TOG_VIEW_LOG checks for requesting commits to populate the log view with calls to view->resize(). ok stsp@

reset committer during rebase and histedit ok op@

call time(3) just once in commit_worktree(), not twice

tog: populate new child view log lines when switching split When switching a child log view from a h- to a v-split, new log lines are empty. Rather than (mis)trace height and (not) call request_log_commits(), let resize_log_view() do the math and request commits if needed to draw new lines. ok op@

do not forget to free(cwd) at the end of cmd_rebase() and cmd_histedit()

fix wrong error return in cmd_commit()

comment tweak

got_repo_pack_fds_close: don't close(0) got_repo_pack_fds_close will try to close all of the GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get a value from got_opentempfd(), which would result in a close(0) if we do not initialize to -1 here.

Correctly track if we sent more data since the last time we looked. With the narrow scope of bytes_sent_cur it would always be zero, and we would always update progress. Found by llvm's scan-build (Dead store). OK stsp

Prevent use-after-free of packed_refs_path in error path. Found by llvm's scan-build. OK stsp

Prevent memory leak when asprintf fails. Found by llvm's scan-build. OK stsp

Make sure got_repo_pack_fds_close() frees a malloc'ed pointer. Found by llvm's scan-build (bad free). OK stsp

If the first readdir() returns NULL err is uninitialized. This can't happen in practice, but llvm's scan-build doesn't know this. OK stsp