Document the force history columns and offsets in case.md instead of a sidecar file

Author: sbryngelson

docs/documentation/case.md

@@ -809,7 +809,31 @@ If `file_per_process` is true, then pre_process, simulation, and post_process mu
 
 - `probe_wrt` activates the output of state variables at coordinates specified by `probe(i)%[x;y,z]`.
 
-- `ib_state_wrt` is used to trigger post-processing of the IB state to be written out as a point mesh in the SILO files. When no IBs are moving, it also triggers force and torque calculation so that those values may be written to the output state files. It also records one line per time step in `D/ib<id>_forces.dat` for each immersed boundary (time step, time, force, torque, velocity, angular velocity, angles, centroid). Records are buffered and written in batches rather than opened per step, and `ib_force_stride` writes only every N-th step for runs long enough that the record itself becomes large.
+- `ib_state_wrt` is used to trigger post-processing of the IB state to be written out as a point mesh in the SILO files. When no IBs are moving, it also triggers force and torque calculation so that those values may be written to the output state files. It also records the force, torque and kinematics of every immersed boundary in a single shared text file, `D/ib_forces.dat`, described below. `ib_force_stride` writes only every N-th step, for runs long enough that the history itself becomes large.
+
+#### Immersed-boundary force history {#sec-ib-force-history}
+
+`D/ib_forces.dat` holds one fixed-width record per body per written step. Its twenty columns are
+
+| Columns | Quantity |
+| ---:    | :---     |
+| 1       | body id (the global `patch_ib` index) |
+| 2       | time |
+| 3–5     | force, x/y/z |
+| 6–8     | torque, x/y/z |
+| 9–11    | velocity, x/y/z |
+| 12–14   | angular velocity, x/y/z |
+| 15–17   | angles about x/y/z |
+| 18–20   | centroid, x/y/z |
+
+The file carries no header line, because every record sits at a computed byte offset and a header would shift them all. Each record is exactly 353 bytes including its newline (`I10` followed by nineteen `1X,ES17.9E3` fields), so the whole file loads with `numpy.loadtxt` and a single body or step can be read without scanning it:
+
+```
+row    = t_step / ib_force_stride
+offset = (row * num_ibs + ib_id - 1) * 353
+```
+
+Rows are written in global body-id order, so the file is byte-identical however the domain is decomposed, and no merge step is needed after a parallel run.
 
 - `output_partial_domain` activates the output of part of the domain specified by `[x,y,z]_output%%beg` and `[x,y,z]_output%%end`.
 This is useful for large domains where only a portion of the domain is of interest.

src/simulation/m_data_output.fpp

@@ -1122,8 +1122,8 @@ contains
     !! 3e5 filesystem metadata operations per step at 1000 ranks holding 100 bodies each.
     !!
     !! Layout: row r = t_step/ib_force_stride holds all num_gbl_ibs bodies in global id order, so
-    !! body g occupies bytes ((r*num_gbl_ibs) + g - 1)*IB_REC_LEN. Column order is documented in
-    !! D/ib_forces.hdr rather than in a header line, which would shift every offset after it.
+    !! body g occupies bytes ((r*num_gbl_ibs) + g - 1)*IB_REC_LEN. The file carries no header line,
+    !! which would shift every offset after it; the columns are listed in docs/documentation/case.md.
     impure subroutine s_write_ib_force_history(t_step)
 
         integer, intent(in)                  :: t_step
@@ -1176,12 +1176,12 @@ contains
 
     end subroutine s_write_ib_force_history
 
-    !> Open the shared history file and drop a sibling naming its columns. Both are done once.
+    !> Open the shared history file. Done once for the run.
     impure subroutine s_open_ib_force_history
 
         character(LEN=path_len + 2*name_len) :: file_loc
         character(LEN=IB_REC_LEN)            :: probe
-        integer                              :: hdr, i
+        integer                              :: i
 
 #ifdef MFC_MPI
         integer :: ierr
@@ -1197,15 +1197,6 @@ contains
         @:PROHIBIT(len_trim(probe) /= IB_REC_BODY, &
                    & "IB force record width disagrees with IB_REC_BODY;  IB_REC_FMT and IB_REC_BODY must be changed together")
 
-        if (proc_rank == 0) then
-            file_loc = trim(case_dir) // '/D/ib_forces.hdr'
-            open (newunit=hdr, file=trim(file_loc), form='formatted', status='replace')
-            write (hdr, '(A)') 'ib_id time Fx Fy Fz Tx Ty Tz vx vy vz wx wy wz ax ay az xc yc zc'
-            write (hdr, '(A,I0,A)') 'record ', IB_REC_LEN, ' bytes; row = t_step/ib_force_stride'
-            write (hdr, '(A)') 'offset(step,body) = (row*num_gbl_ibs + ib_id - 1)*record'
-            close (hdr)
-        end if
-
         file_loc = trim(case_dir) // '/D/ib_forces.dat'
 #ifdef MFC_MPI
         ! MPI_MODE_CREATE does not truncate, so a shorter run following a longer one in the same